From fe21aab340b60719555e1b917b8ea734df515441 Mon Sep 17 00:00:00 2001 From: Junjie Qi Date: Mon, 2 Dec 2024 17:23:45 -0800 Subject: [PATCH 01/10] [1/n] pin the dependecies version for x86_64 --- .github/actions/build_cmake/action.yml | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/.github/actions/build_cmake/action.yml b/.github/actions/build_cmake/action.yml index b1a17d58b8..a3b2002713 100644 --- a/.github/actions/build_cmake/action.yml +++ b/.github/actions/build_cmake/action.yml @@ -32,7 +32,7 @@ runs: conda update -y -q conda echo "$CONDA/bin" >> $GITHUB_PATH - conda install -y -q python=3.11 cmake make swig "numpy<2" scipy pytest gflags + conda install -y -q python=3.11 cmake=3.26 make=4.2 swig=4.0 "numpy<2" scipy=1.14 pytest=7.4 gflags=2.2 # install base packages for ARM64 if [ "${{ runner.arch }}" = "ARM64" ]; then @@ -42,7 +42,7 @@ runs: # install base packages for X86_64 if [ "${{ runner.arch }}" = "X64" ]; then # TODO: merge this with ARM64 - conda install -y -q -c conda-forge gxx_linux-64 sysroot_linux-64 + conda install -y -q -c conda-forge gxx_linux-64=14.2 sysroot_linux-64=2.17 conda install -y -q mkl=2023 mkl-devel=2023 fi @@ -58,14 +58,13 @@ runs: fi # install test packages - conda install -y pytest if [ "${{ inputs.rocm }}" = "ON" ]; then : # skip torch install via conda, we need to install via pip to get # ROCm-enabled version until it's supported in conda by PyTorch elif [ "${{ inputs.gpu }}" = "ON" ]; then conda install -y -q "pytorch<2.5" pytorch-cuda=12.4 -c pytorch -c nvidia/label/cuda-12.4.0 else - conda install -y -q pytorch -c pytorch + conda install -y -q pytorch=2.5 -c pytorch fi - name: ROCm - Install dependencies if: inputs.rocm == 'ON' From 7cced896498dbddbc9afe637107f6baf747e8365 Mon Sep 17 00:00:00 2001 From: Junjie Qi Date: Mon, 2 Dec 2024 22:03:49 -0800 Subject: [PATCH 02/10] add pytest back --- .github/actions/build_cmake/action.yml | 1 + 1 file changed, 1 insertion(+) diff --git a/.github/actions/build_cmake/action.yml b/.github/actions/build_cmake/action.yml index a3b2002713..72c19693e6 100644 --- a/.github/actions/build_cmake/action.yml +++ b/.github/actions/build_cmake/action.yml @@ -58,6 +58,7 @@ runs: fi # install test packages + conda install -y pytest if [ "${{ inputs.rocm }}" = "ON" ]; then : # skip torch install via conda, we need to install via pip to get # ROCm-enabled version until it's supported in conda by PyTorch From 3cde385a3e3db8dd96f803d608608dca815d1a44 Mon Sep 17 00:00:00 2001 From: Junjie Qi Date: Mon, 2 Dec 2024 22:58:08 -0800 Subject: [PATCH 03/10] arm 64 sve pytorch support --- .github/actions/build_cmake/action.yml | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/.github/actions/build_cmake/action.yml b/.github/actions/build_cmake/action.yml index 72c19693e6..d398991f2a 100644 --- a/.github/actions/build_cmake/action.yml +++ b/.github/actions/build_cmake/action.yml @@ -58,14 +58,13 @@ runs: fi # install test packages - conda install -y pytest if [ "${{ inputs.rocm }}" = "ON" ]; then : # skip torch install via conda, we need to install via pip to get # ROCm-enabled version until it's supported in conda by PyTorch elif [ "${{ inputs.gpu }}" = "ON" ]; then conda install -y -q "pytorch<2.5" pytorch-cuda=12.4 -c pytorch -c nvidia/label/cuda-12.4.0 else - conda install -y -q pytorch=2.5 -c pytorch + conda install -y -q "pytorch<2.5" -c pytorch fi - name: ROCm - Install dependencies if: inputs.rocm == 'ON' From 7210748e4ec66f80133c2adf6580c2184fc072a0 Mon Sep 17 00:00:00 2001 From: Junjie Qi Date: Tue, 3 Dec 2024 14:02:19 -0800 Subject: [PATCH 04/10] try to fix cmake version --- .github/actions/build_cmake/action.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/actions/build_cmake/action.yml b/.github/actions/build_cmake/action.yml index d398991f2a..7b7bb57c3d 100644 --- a/.github/actions/build_cmake/action.yml +++ b/.github/actions/build_cmake/action.yml @@ -32,7 +32,7 @@ runs: conda update -y -q conda echo "$CONDA/bin" >> $GITHUB_PATH - conda install -y -q python=3.11 cmake=3.26 make=4.2 swig=4.0 "numpy<2" scipy=1.14 pytest=7.4 gflags=2.2 + conda install -y -q python=3.11 cmake>=3.26.4 make=4.2 swig=4.0 "numpy<2" scipy=1.14 pytest=7.4 gflags=2.2 # install base packages for ARM64 if [ "${{ runner.arch }}" = "ARM64" ]; then From 13df2753c5d348d8ccf51ec84ec294e06361f1e2 Mon Sep 17 00:00:00 2001 From: Junjie Qi Date: Tue, 3 Dec 2024 14:04:00 -0800 Subject: [PATCH 05/10] try to fix cmake version --- .github/actions/build_cmake/action.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/actions/build_cmake/action.yml b/.github/actions/build_cmake/action.yml index 7b7bb57c3d..b93c813649 100644 --- a/.github/actions/build_cmake/action.yml +++ b/.github/actions/build_cmake/action.yml @@ -32,7 +32,7 @@ runs: conda update -y -q conda echo "$CONDA/bin" >> $GITHUB_PATH - conda install -y -q python=3.11 cmake>=3.26.4 make=4.2 swig=4.0 "numpy<2" scipy=1.14 pytest=7.4 gflags=2.2 + conda install -y -q python=3.11 "cmake>=3.26.4" make=4.2 swig=4.0 "numpy<2" scipy=1.14 pytest=7.4 gflags=2.2 # install base packages for ARM64 if [ "${{ runner.arch }}" = "ARM64" ]; then From e79c9f6317cb1e326a2afa4b3ff3e8a9c8476023 Mon Sep 17 00:00:00 2001 From: Junjie Qi Date: Tue, 3 Dec 2024 15:49:43 -0800 Subject: [PATCH 06/10] try to fix cmake version --- .github/actions/build_cmake/action.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/actions/build_cmake/action.yml b/.github/actions/build_cmake/action.yml index b93c813649..6e3d877aa6 100644 --- a/.github/actions/build_cmake/action.yml +++ b/.github/actions/build_cmake/action.yml @@ -54,7 +54,7 @@ runs: conda install -y -q cuda-toolkit -c "nvidia/label/cuda-12.4.0" # and CUDA from cuVS channel for cuVS builds elif [ "${{ inputs.cuvs }}" = "ON" ]; then - conda install -y -q libcuvs=24.08 cuda-version=12.4 cuda-toolkit -c rapidsai -c conda-forge -c "nvidia/label/cuda-12.4.0" + conda install -y -q libcuvs=24.08 cuda-version=12.4.1 cuda-toolkit -c rapidsai -c conda-forge -c "nvidia/label/cuda-12.4.0" fi # install test packages From dd988fa2e1f70c68a7ee6540d916d57939cec754 Mon Sep 17 00:00:00 2001 From: Junjie Qi Date: Tue, 3 Dec 2024 16:26:40 -0800 Subject: [PATCH 07/10] try to fix cmake version --- .github/actions/build_cmake/action.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/.github/actions/build_cmake/action.yml b/.github/actions/build_cmake/action.yml index 6e3d877aa6..9c6f168753 100644 --- a/.github/actions/build_cmake/action.yml +++ b/.github/actions/build_cmake/action.yml @@ -32,7 +32,7 @@ runs: conda update -y -q conda echo "$CONDA/bin" >> $GITHUB_PATH - conda install -y -q python=3.11 "cmake>=3.26.4" make=4.2 swig=4.0 "numpy<2" scipy=1.14 pytest=7.4 gflags=2.2 + conda install -y -q python=3.11 cmake make swig=4.0 "numpy<2" scipy=1.14 pytest=7.4 gflags=2.2 # install base packages for ARM64 if [ "${{ runner.arch }}" = "ARM64" ]; then @@ -54,7 +54,7 @@ runs: conda install -y -q cuda-toolkit -c "nvidia/label/cuda-12.4.0" # and CUDA from cuVS channel for cuVS builds elif [ "${{ inputs.cuvs }}" = "ON" ]; then - conda install -y -q libcuvs=24.08 cuda-version=12.4.1 cuda-toolkit -c rapidsai -c conda-forge -c "nvidia/label/cuda-12.4.0" + conda install -y -q libcuvs=24.08 cuda-version=12.4 cuda-toolkit -c rapidsai -c conda-forge -c "nvidia/label/cuda-12.4.0" fi # install test packages From 30aac092c3720c36e7cf57890cc92ac5ee1ac5d8 Mon Sep 17 00:00:00 2001 From: Junjie Qi Date: Tue, 3 Dec 2024 17:38:28 -0800 Subject: [PATCH 08/10] try to fix cmake version --- .github/actions/build_cmake/action.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/.github/actions/build_cmake/action.yml b/.github/actions/build_cmake/action.yml index 9c6f168753..9fce19c433 100644 --- a/.github/actions/build_cmake/action.yml +++ b/.github/actions/build_cmake/action.yml @@ -32,7 +32,7 @@ runs: conda update -y -q conda echo "$CONDA/bin" >> $GITHUB_PATH - conda install -y -q python=3.11 cmake make swig=4.0 "numpy<2" scipy=1.14 pytest=7.4 gflags=2.2 + conda install -y -q python=3.11 cmake=3.26 make=4.2 swig=4.0 "numpy<2" scipy=1.14 pytest=7.4 gflags=2.2 # install base packages for ARM64 if [ "${{ runner.arch }}" = "ARM64" ]; then @@ -54,7 +54,7 @@ runs: conda install -y -q cuda-toolkit -c "nvidia/label/cuda-12.4.0" # and CUDA from cuVS channel for cuVS builds elif [ "${{ inputs.cuvs }}" = "ON" ]; then - conda install -y -q libcuvs=24.08 cuda-version=12.4 cuda-toolkit -c rapidsai -c conda-forge -c "nvidia/label/cuda-12.4.0" + conda install -y -q libcuvs=24.08 cuda-version=12.4 cuda-toolkit gxx_linux-64=12.4 -c rapidsai -c conda-forge -c "nvidia/label/cuda-12.4.0" fi # install test packages From f4ace217fcf2b18d3367ace816aca5628668868e Mon Sep 17 00:00:00 2001 From: Junjie Qi Date: Tue, 3 Dec 2024 21:12:24 -0800 Subject: [PATCH 09/10] pin cuda version --- .github/actions/build_cmake/action.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/.github/actions/build_cmake/action.yml b/.github/actions/build_cmake/action.yml index 9fce19c433..1c2b1522e0 100644 --- a/.github/actions/build_cmake/action.yml +++ b/.github/actions/build_cmake/action.yml @@ -51,10 +51,10 @@ runs: : # regular CUDA for GPU builds elif [ "${{ inputs.gpu }}" = "ON" ] && [ "${{ inputs.cuvs }}" = "OFF" ]; then - conda install -y -q cuda-toolkit -c "nvidia/label/cuda-12.4.0" + conda install -y -q cuda-toolkit=12.4 -c "nvidia/label/cuda-12.4.0" # and CUDA from cuVS channel for cuVS builds elif [ "${{ inputs.cuvs }}" = "ON" ]; then - conda install -y -q libcuvs=24.08 cuda-version=12.4 cuda-toolkit gxx_linux-64=12.4 -c rapidsai -c conda-forge -c "nvidia/label/cuda-12.4.0" + conda install -y -q libcuvs=24.08 cuda-version=12.4 cuda-toolkit=12.4 gxx_linux-64=12.4 -c rapidsai -c conda-forge -c "nvidia/label/cuda-12.4.0" fi # install test packages From 7e5be631791ad180684dc3e3b54251b0c47c3602 Mon Sep 17 00:00:00 2001 From: Junjie Qi Date: Tue, 3 Dec 2024 22:13:51 -0800 Subject: [PATCH 10/10] unpin cuda-toolkit version first --- .github/actions/build_cmake/action.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/.github/actions/build_cmake/action.yml b/.github/actions/build_cmake/action.yml index 1c2b1522e0..9fce19c433 100644 --- a/.github/actions/build_cmake/action.yml +++ b/.github/actions/build_cmake/action.yml @@ -51,10 +51,10 @@ runs: : # regular CUDA for GPU builds elif [ "${{ inputs.gpu }}" = "ON" ] && [ "${{ inputs.cuvs }}" = "OFF" ]; then - conda install -y -q cuda-toolkit=12.4 -c "nvidia/label/cuda-12.4.0" + conda install -y -q cuda-toolkit -c "nvidia/label/cuda-12.4.0" # and CUDA from cuVS channel for cuVS builds elif [ "${{ inputs.cuvs }}" = "ON" ]; then - conda install -y -q libcuvs=24.08 cuda-version=12.4 cuda-toolkit=12.4 gxx_linux-64=12.4 -c rapidsai -c conda-forge -c "nvidia/label/cuda-12.4.0" + conda install -y -q libcuvs=24.08 cuda-version=12.4 cuda-toolkit gxx_linux-64=12.4 -c rapidsai -c conda-forge -c "nvidia/label/cuda-12.4.0" fi # install test packages