From 18bc46b3b4ca63e7d7cded3264485084185449d3 Mon Sep 17 00:00:00 2001 From: Robin Date: Wed, 20 Aug 2025 15:25:59 +0800 Subject: [PATCH 1/4] Update build-wheels-cuda.yaml --- .github/workflows/build-wheels-cuda.yaml | 19 +++++-------------- 1 file changed, 5 insertions(+), 14 deletions(-) diff --git a/.github/workflows/build-wheels-cuda.yaml b/.github/workflows/build-wheels-cuda.yaml index 07b30cfc0..629517697 100644 --- a/.github/workflows/build-wheels-cuda.yaml +++ b/.github/workflows/build-wheels-cuda.yaml @@ -8,7 +8,7 @@ permissions: jobs: define_matrix: name: Define Build Matrix - runs-on: ubuntu-22.04 + runs-on: ubuntu-latest outputs: matrix: ${{ steps.set-matrix.outputs.matrix }} defaults: @@ -20,9 +20,9 @@ jobs: id: set-matrix run: | $matrix = @{ - 'os' = @('ubuntu-22.04') #, 'windows-2022') - 'pyver' = @("3.9", "3.10", "3.11", "3.12") - 'cuda' = @("12.1.1", "12.2.2", "12.3.2", "12.4.1") #, "12.5.1", "12.6.1") + 'os' = @('ubuntu-latest', 'windows-2022') + 'pyver' = @("3.12", "3.13") + 'cuda' = @("12.6.1") 'releasetag' = @("basic") } @@ -112,17 +112,8 @@ jobs: $env:LD_LIBRARY_PATH = $env:CONDA_PREFIX + '/lib:' + $env:LD_LIBRARY_PATH } $env:VERBOSE = '1' - $env:CMAKE_ARGS = '-DGGML_CUDA=on -DCMAKE_CUDA_ARCHITECTURES=all' + $env:CMAKE_ARGS = '-DGGML_CUDA=on -DCMAKE_CUDA_ARCHITECTURES=89 -DGGML_CUDA_FA_ALL_QUANTS=ON' $env:CMAKE_ARGS = "-DGGML_CUDA_FORCE_MMQ=ON $env:CMAKE_ARGS" - # if ($env:AVXVER -eq 'AVX') { - $env:CMAKE_ARGS = $env:CMAKE_ARGS + ' -DGGML_AVX2=off -DGGML_FMA=off -DGGML_F16C=off' - # } - # if ($env:AVXVER -eq 'AVX512') { - # $env:CMAKE_ARGS = $env:CMAKE_ARGS + ' -DGGML_AVX512=on' - # } - # if ($env:AVXVER -eq 'basic') { - # $env:CMAKE_ARGS = $env:CMAKE_ARGS + ' -DGGML_AVX=off -DGGML_AVX2=off -DGGML_FMA=off -DGGML_F16C=off' - # } python -m build --wheel # write the build tag to the output Write-Output "CUDA_VERSION=$cudaVersion" >> $env:GITHUB_ENV From aa0c81e1678ad0a7007e93ec8749df0003560321 Mon Sep 17 00:00:00 2001 From: Robin Date: Wed, 20 Aug 2025 15:31:02 +0800 Subject: [PATCH 2/4] Update build-wheels-cuda.yaml --- .github/workflows/build-wheels-cuda.yaml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/workflows/build-wheels-cuda.yaml b/.github/workflows/build-wheels-cuda.yaml index 629517697..16fe0b935 100644 --- a/.github/workflows/build-wheels-cuda.yaml +++ b/.github/workflows/build-wheels-cuda.yaml @@ -21,7 +21,7 @@ jobs: run: | $matrix = @{ 'os' = @('ubuntu-latest', 'windows-2022') - 'pyver' = @("3.12", "3.13") + 'pyver' = @("3.12") 'cuda' = @("12.6.1") 'releasetag' = @("basic") } From 76574f77f9cb25dd2a2e6ddfd411112287e36ec6 Mon Sep 17 00:00:00 2001 From: Robin Date: Wed, 20 Aug 2025 15:33:39 +0800 Subject: [PATCH 3/4] Update build-wheels-cuda.yaml --- .github/workflows/build-wheels-cuda.yaml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/.github/workflows/build-wheels-cuda.yaml b/.github/workflows/build-wheels-cuda.yaml index 16fe0b935..f69e872c3 100644 --- a/.github/workflows/build-wheels-cuda.yaml +++ b/.github/workflows/build-wheels-cuda.yaml @@ -8,7 +8,7 @@ permissions: jobs: define_matrix: name: Define Build Matrix - runs-on: ubuntu-latest + runs-on: ubuntu-22.04 outputs: matrix: ${{ steps.set-matrix.outputs.matrix }} defaults: @@ -20,7 +20,7 @@ jobs: id: set-matrix run: | $matrix = @{ - 'os' = @('ubuntu-latest', 'windows-2022') + 'os' = @('ubuntu-22.04', 'windows-2022') 'pyver' = @("3.12") 'cuda' = @("12.6.1") 'releasetag' = @("basic") From b95ad94a9a6b3eeb8bf9b7dd19e3e5c7e60b2cc4 Mon Sep 17 00:00:00 2001 From: Robin Date: Thu, 21 Aug 2025 01:20:13 +0800 Subject: [PATCH 4/4] Update build-wheels-cuda.yaml --- .github/workflows/build-wheels-cuda.yaml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/workflows/build-wheels-cuda.yaml b/.github/workflows/build-wheels-cuda.yaml index f69e872c3..1bdf068a0 100644 --- a/.github/workflows/build-wheels-cuda.yaml +++ b/.github/workflows/build-wheels-cuda.yaml @@ -22,7 +22,7 @@ jobs: $matrix = @{ 'os' = @('ubuntu-22.04', 'windows-2022') 'pyver' = @("3.12") - 'cuda' = @("12.6.1") + 'cuda' = @("12.4.1") 'releasetag' = @("basic") }