Skip to content

fix(ci): reduce CUDA binary wheel size only including cubins for current arches and one PTX target for forward compatibility #2158

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
abetlen merged 2 commits into from
Mar 25, 2026
+6 −3
Merged

fix(ci): reduce CUDA binary wheel size only including cubins for current arches and one PTX target for forward compatibility #2158

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
94fd771
fix(ci): shrink CUDA wheel fatbins
abetlen Mar 25, 2026
36d1675
docs: update changelog for cuda wheel size fix
abetlen Mar 25, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 4 additions & 3 deletions .github/workflows/build-wheels-cuda.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -153,9 +153,10 @@ jobs:
}
$cudaTagVersion = $nvccVersion.Replace('.','')
$env:VERBOSE = '1'
# Keep a portable SM set, including sm_70, instead of CMake's `all`,
# which now pulls in future targets the hosted-runner toolchains cannot assemble.
$env:CMAKE_ARGS = "-DGGML_CUDA_FORCE_MMQ=ON -DGGML_CUDA=on -DCMAKE_CUDA_ARCHITECTURES=70;75;80;86;89;90 -DCMAKE_CUDA_FLAGS=--allow-unsupported-compiler $env:CMAKE_ARGS"
# Build real cubins for the supported GPUs, including sm_70, and keep
# one forward-compatible PTX target instead of embedding PTX for every
# SM. This keeps the wheel under GitHub's 2 GiB release-asset limit.
$env:CMAKE_ARGS = "-DGGML_CUDA_FORCE_MMQ=ON -DGGML_CUDA=on -DCMAKE_CUDA_ARCHITECTURES=70-real;75-real;80-real;86-real;89-real;90-real;90-virtual -DCMAKE_CUDA_FLAGS=--allow-unsupported-compiler $env:CMAKE_ARGS"
# if ($env:AVXVER -eq 'AVX') {
$env:CMAKE_ARGS = $env:CMAKE_ARGS + ' -DGGML_AVX2=off -DGGML_FMA=off -DGGML_F16C=off'
# }
Expand Down
2 changes: 2 additions & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0

## [Unreleased]

- fix(ci): Shrink CUDA wheel fatbins so CUDA releases stay under GitHub's asset size limit by @abetlen in #2158

## [0.3.18]

- feat: Expose `attention_type` in `Llama.__init__` for non-causal embedding models by @jamesbiederbeck in #2143
Expand Down
Loading