diff --git a/.github/actions/inductor-xpu-e2e-test/action.yml b/.github/actions/inductor-xpu-e2e-test/action.yml index 9f2828c3b..2164b9eee 100644 --- a/.github/actions/inductor-xpu-e2e-test/action.yml +++ b/.github/actions/inductor-xpu-e2e-test/action.yml @@ -50,7 +50,7 @@ runs: if: ${{ inputs.env_prepare }} shell: bash run: | - source activate e2e_ci + source ${HOME}/mengfeil/myvenv-e2e/bin/activate source .github/scripts/env.sh ${{ inputs.pytorch }} if [[ ${{ inputs.suite }} == *"torchbench"* ]]; then if [ "${{ inputs.pytorch }}" != "nightly_wheel" ]; then @@ -89,11 +89,9 @@ runs: - name: E2E Test (${{ inputs.suite }} ${{ inputs.dt }} ${{ inputs.mode }} ${{ inputs.scenario }}) env: HUGGING_FACE_HUB_TOKEN: ${{ inputs.hf_token }} - NEOReadDebugKeys: 1 - DisableScratchPages: 1 shell: bash run: | - source activate e2e_ci + source ${HOME}/mengfeil/myvenv-e2e/bin/activate source .github/scripts/env.sh ${{ inputs.pytorch }} cp .github/scripts/inductor_xpu_test.sh ../pytorch cd ../pytorch diff --git a/.github/workflows/_linux_ut.yml b/.github/workflows/_linux_ut.yml index 953e9b1bc..65932365b 100644 --- a/.github/workflows/_linux_ut.yml +++ b/.github/workflows/_linux_ut.yml @@ -50,20 +50,15 @@ jobs: Torch-XPU-UT-Tests: runs-on: ${{ inputs.runner }} timeout-minutes: 9000 - env: - NEOReadDebugKeys: 1 - DisableScratchPages: 1 steps: - name: Checkout torch-xpu-ops uses: actions/checkout@v4 - name: Prepare Stock Pytorch run: | pwd - which conda && conda clean -ay - conda remove --all -y -n xpu_op_${ZE_AFFINITY_MASK} || \ - rm -rf $(dirname ${CONDA_EXE})/../envs/xpu_op_${ZE_AFFINITY_MASK} - conda create -n xpu_op_${ZE_AFFINITY_MASK} python=${{ inputs.python }} cmake ninja -y - source activate xpu_op_${ZE_AFFINITY_MASK} + rm -rf ${HOME}/mengfeil/myvenv-ut + /usr/bin/python3.10 -m venv ${HOME}/mengfeil/myvenv-ut + source ${HOME}/mengfeil/myvenv-ut/bin/activate cd ../ && rm -rf pytorch pip install requests git clone https://github.com/pytorch/pytorch pytorch @@ -83,7 +78,7 @@ jobs: fi - name: Triton Installation run: | - source activate xpu_op_${ZE_AFFINITY_MASK} + source ${HOME}/mengfeil/myvenv-ut/bin/activate cd ../pytorch TRITON_REPO="https://github.com/intel/intel-xpu-backend-for-triton" if [ -z ${{ inputs.triton }} ]; then @@ -97,7 +92,7 @@ jobs: fi - name: Build Pytorch XPU run: | - source activate xpu_op_${ZE_AFFINITY_MASK} + source ${HOME}/mengfeil/myvenv-ut/bin/activate source .github/scripts/env.sh ${{ inputs.pytorch }} if [[ ${{ inputs.abi }} == '0' ]]; then export _GLIBCXX_USE_CXX11_ABI=0 @@ -127,7 +122,7 @@ jobs: pip install -r .ci/docker/requirements-ci.txt - name: Torch Config run: | - source activate xpu_op_${ZE_AFFINITY_MASK} + source ${HOME}/mengfeil/myvenv-ut/bin/activate source .github/scripts/env.sh ${{ inputs.pytorch }} python -c "import torch; print(torch.__config__.show())" python -c "import torch; print(torch.__config__.parallel_info())" @@ -141,10 +136,10 @@ jobs: - name: Run XPU OP Examples if: contains(inputs.ut, 'op_regression') || github.event_name == 'schedule' run: | + source ${HOME}/mengfeil/myvenv-ut/bin/activate cd ${{ github.workspace }} xpu-smi discovery source .github/scripts/env.sh ${{ inputs.pytorch }} - source activate xpu_op_${ZE_AFFINITY_MASK} cd ${{ github.workspace }} cd test/regressions pip install pytest @@ -152,10 +147,10 @@ jobs: - name: Run XPU OP Regressions test on device 1 if: contains(inputs.ut, 'op_regression_dev1') || github.event_name == 'schedule' run: | + source ${HOME}/mengfeil/myvenv-ut/bin/activate cd ${{ github.workspace }} xpu-smi discovery source .github/scripts/env.sh ${{ inputs.pytorch }} - source activate xpu_op_${ZE_AFFINITY_MASK} export ZE_AFFINITY_MASK_OLD=${ZE_AFFINITY_MASK} unset ZE_AFFINITY_MASK cd ${{ github.workspace }} @@ -167,7 +162,7 @@ jobs: if: contains(inputs.ut, 'op_extended') || github.event_name == 'schedule' run: | source .github/scripts/env.sh ${{ inputs.pytorch }} - source activate xpu_op_${ZE_AFFINITY_MASK} + source ${HOME}/mengfeil/myvenv-ut/bin/activate export PYTORCH_TEST_WITH_SLOW=1 cd ../pytorch/third_party/torch-xpu-ops/test/xpu/extended/ timeout 10000 python run_test_with_skip.py || true @@ -175,7 +170,7 @@ jobs: if: contains(inputs.ut, 'op_ut') || github.event_name == 'schedule' run: | source .github/scripts/env.sh ${{ inputs.pytorch }} - source activate xpu_op_${ZE_AFFINITY_MASK} + source ${HOME}/mengfeil/myvenv-ut/bin/activate export PYTORCH_ENABLE_XPU_FALLBACK=1 export PYTORCH_TEST_WITH_SLOW=1 cd ../pytorch/third_party/torch-xpu-ops/test/xpu @@ -189,7 +184,7 @@ jobs: if: contains(inputs.ut, 'torch_xpu') || github.event_name == 'schedule' run: | source .github/scripts/env.sh ${{ inputs.pytorch }} - source activate xpu_op_${ZE_AFFINITY_MASK} + source ${HOME}/mengfeil/myvenv-ut/bin/activate cd ../pytorch TEST_REPORTS_DIR=$(pwd)/test/test-reports rm -rf "$TEST_REPORTS_DIR" && mkdir -p "$TEST_REPORTS_DIR" diff --git a/.github/workflows/nightly_ondemand_whl.yml b/.github/workflows/nightly_ondemand_whl.yml index 889d8614a..4815f15d6 100644 --- a/.github/workflows/nightly_ondemand_whl.yml +++ b/.github/workflows/nightly_ondemand_whl.yml @@ -62,10 +62,10 @@ jobs: ut: ${{ github.event_name == 'schedule' && 'op_regression,op_regression_dev1,op_extended,op_ut,torch_xpu' || inputs.ut }} python: ${{ github.event_name == 'schedule' && '3.10' || inputs.python }} pytorch: nightly_wheel - runner: pvc_rolling + runner: e2e_internal Linux-Nightly-Ondemand-E2E-WHL-Tests: - runs-on: pvc_rolling + runs-on: e2e_internal # Don't run on forked repos if: github.repository_owner == 'intel' timeout-minutes: 36000 @@ -73,8 +73,6 @@ jobs: pytorch: ${{ github.event_name == 'schedule' && 'nightly' || inputs.pytorch }} ut: ${{ github.event_name == 'schedule' && 'op_regression,op_regression_dev1,op_extended,op_ut,torch_xpu' || inputs.ut }} python: ${{ github.event_name == 'schedule' && '3.10' || inputs.python }} - NEOReadDebugKeys: 1 - DisableScratchPages: 1 outputs: TORCH_BRANCH_ID: ${{ steps.installed.outputs.TORCH_BRANCH_ID }} TORCH_COMMIT_ID: ${{ steps.installed.outputs.TORCH_COMMIT_ID }} @@ -96,17 +94,16 @@ jobs: uses: actions/checkout@v4 - name: Prepare Conda ENV run: | - which conda && conda clean -ay - conda remove --all -y -n e2e_ci || rm -rf $(dirname ${CONDA_EXE})/../envs/e2e_ci - conda create -n e2e_ci python=${{ env.python }} cmake ninja -y - source activate e2e_ci + rm -rf ${HOME}/mengfeil/myvenv-e2e + /usr/bin/python3.10 -m venv ${HOME}/mengfeil/myvenv-e2e + source ${HOME}/mengfeil/myvenv-e2e/bin/activate # pip install mkl-static==2025.0.1 mkl-include==2025.0.1 pip install pandas scipy tqdm - name: Prepare Stock Pytorch id: installed run: | pwd - source activate e2e_ci + source ${HOME}/mengfeil/myvenv-e2e/bin/activate pip install torch torchvision torchaudio --pre --index-url https://download.pytorch.org/whl/nightly/xpu echo "TORCH_BRANCH_ID=$(python -c 'import torch; print(torch.__version__)')" |tee -a "${GITHUB_OUTPUT}" >> "${GITHUB_ENV}" TORCH_COMMIT_ID=$(python -c 'import torch; print(torch.version.git_version)') @@ -128,7 +125,7 @@ jobs: - name: Identify pinned versions id: pinned run: | - source activate e2e_ci + source ${HOME}/mengfeil/myvenv-e2e/bin/activate echo "TORCHVISION_COMMIT_ID=$(python -c 'import torchvision; print(torchvision.version.git_version)')" |tee -a "${GITHUB_OUTPUT}" >> "${GITHUB_ENV}" echo "TORCHAUDIO_COMMIT_ID=$(python -c 'import torchaudio; print(torchaudio.version.git_version)')" |tee -a "${GITHUB_OUTPUT}" >> "${GITHUB_ENV}" echo "TRITON_COMMIT_ID=$(python -c 'import triton; print(triton.__version__)')" |tee -a "${GITHUB_OUTPUT}" >> "${GITHUB_ENV}"