Install E2E dependencies for Windows (#4496)

pbchekin · web-flow · commit bf7a0317a66f · 2025-06-13T09:14:05.000-07:00
Updating reusable actions (cache save/load, install-dependency) to work
on Windows. Note there are no actual e2e tests in the workflow.

Signed-off-by: Pavel Chekin &lt;pavel.chekin@intel.com&gt;
diff --git a/.github/actions/install-dependency/action.yml b/.github/actions/install-dependency/action.yml
@@ -16,6 +16,9 @@ inputs:
   extra-cache-key:
     description: Cache key suffix
     default: ""
+  workspace:
+    description: Workspace directory
+    default: ""
 runs:
   using: "composite"
   steps:
@@ -30,6 +33,9 @@ runs:
     - name: Identify Python version
       shell: bash
       run: |
+        if [[ -f .venv/Scripts/activate ]]; then
+          source .venv/Scripts/activate
+        fi
         if [[ -z "$PYTHON_VERSION" ]]; then
           PYTHON_VERSION="$(python -c 'import sys; print(f"{sys.version_info[0]}.{ sys.version_info[1]}")')"
           echo "PYTHON_VERSION=$PYTHON_VERSION" >> "$GITHUB_ENV"
@@ -40,46 +46,93 @@ runs:
       run: |
         DEPENDENCY_CACHE_KEY="${{ inputs.package }}-$PYTHON_VERSION-$(echo ${{ steps.commit-id.outputs.commit_id }} ${{ inputs.extra-cache-key }} | sha256sum - | cut -d\  -f1)"
         echo "DEPENDENCY_CACHE_KEY=$DEPENDENCY_CACHE_KEY" | tee -a "$GITHUB_ENV"
+        if [[ "${{ inputs.workspace }}" ]]; then
+          echo "PACKAGE_PATH=${{ inputs.workspace }}/${{ inputs.package }}" | tee -a "$GITHUB_ENV"
+        else
+          echo "PACKAGE_PATH=${{ inputs.package }}" | tee -a "$GITHUB_ENV"
+        fi
 
     - name: Try to load package from a cache
       id: cache-load
       uses: ./.github/actions/load
       with:
-        path: ${{ inputs.package }}-${{ steps.commit-id.outputs.commit_id }}
+        path: ${{ env.PACKAGE_PATH }}
         key: ${{ env.DEPENDENCY_CACHE_KEY }}
 
     - name: Clone package repo
-      if: ${{ steps.cache-load.outputs.status == 'miss' }}
+      if: steps.cache-load.outputs.status == 'miss'
       uses: actions/checkout@v4
       with:
         repository: ${{ inputs.repository }}
         ref: ${{ steps.commit-id.outputs.commit_id }}
         submodules: recursive
-        path: ${{ inputs.package }}-${{ steps.commit-id.outputs.commit_id }}
+        path: ${{ inputs.package }}
 
-    - name: Build package wheels
-      if: ${{ steps.cache-load.outputs.status == 'miss' }}
+    - name: Move package repo to the workspace
+      if: inputs.workspace != '' && steps.cache-load.outputs.status == 'miss'
+      shell: bash
+      run: |
+        rm -rf ${{ env.PACKAGE_PATH }}
+        mv -T ${{ inputs.package }} ${{ env.PACKAGE_PATH }}
+
+    - name: Build package wheels (Linux)
+      if: runner.os == 'Linux' && steps.cache-load.outputs.status == 'miss'
       shell: bash
       run: |
         source /opt/intel/oneapi/setvars.sh
-        cd ${{ inputs.package }}-${{ steps.commit-id.outputs.commit_id }}
+
+        cd ${{ env.PACKAGE_PATH }}
         if [[ -f setup.py ]]; then
           python setup.py bdist_wheel
         elif [[ -f pyproject.toml ]]; then
           pip install build
           python -m build
         else
-          echo "Error: setup.py and pyproject.toml not found"
+          echo Error: setup.py and pyproject.toml not found
           exit 1
         fi
 
-    - name: Install package wheels
+    - name: Build package wheels (Windows)
+      if: runner.os == 'Windows' && steps.cache-load.outputs.status == 'miss'
+      shell: pwsh
+      run: |
+        .venv\Scripts\activate.ps1
+        Invoke-BatchFile "C:\Program Files (x86)\Intel\oneAPI\setvars.bat"
+
+        bash -c "
+        cd ${{ env.PACKAGE_PATH }}
+        if [[ -f setup.py ]]; then
+          python setup.py bdist_wheel
+        elif [[ -f pyproject.toml ]]; then
+          pip install build
+          python -m build
+        else
+          echo Error: setup.py and pyproject.toml not found
+          exit 1
+        fi
+        "
+
+    - name: Install package wheels (Linux)
+      if: runner.os == 'Linux'
       shell: bash
       run: |
         source /opt/intel/oneapi/setvars.sh
-        pip install ${{ inputs.package }}-${{ steps.commit-id.outputs.commit_id }}/dist/*.whl
+
+        pip install ${{ env.PACKAGE_PATH }}/dist/*.whl
         python -c "import ${{ inputs.package }}; print(${{ inputs.package }}.__version__)"
 
+    - name: Install package wheels (Windows)
+      if: runner.os == 'Windows'
+      shell: pwsh
+      run: |
+        .venv\Scripts\activate.ps1
+        Invoke-BatchFile "C:\Program Files (x86)\Intel\oneAPI\setvars.bat"
+
+        bash -c "
+        pip install ${{ env.PACKAGE_PATH }}/dist/*.whl
+        python -c 'import ${{ inputs.package }}; print(${{ inputs.package }}.__version__)'
+        "
+
     - name: Save package to a cache
       if: ${{ steps.cache-load.outputs.status == 'miss' }}
       uses: ./.github/actions/save
diff --git a/.github/actions/load/action.yml b/.github/actions/load/action.yml
@@ -34,16 +34,23 @@ runs:
       id: load
       shell: bash
       run: |
+        if [[ ! -d ${{ inputs.root }} ]]; then
+          echo "${{ inputs.root }} does not exist, cache disabled"
+          echo "status=miss" >> $GITHUB_OUTPUT
+          exit 0
+        fi
+
         ITEM_PATH="${{ inputs.root }}/${{ inputs.key }}"
         echo "dest=$ITEM_PATH" >> $GITHUB_OUTPUT
         if [[ -d ${{ inputs.path }} ]]; then
           echo "Directory ${{ inputs.path }} already exists and will be removed"
           rm -rf ${{ inputs.path }}
         fi
+
         if [[ ${{ inputs.enabled == 'true' }} && -d $ITEM_PATH ]]; then
           echo "Cache hit for ${{ inputs.key }}"
           echo "status=hit" >> $GITHUB_OUTPUT
-          if [[ ${{ inputs.symlink }} == true ]]; then
+          if [[ ${{ inputs.symlink }} == true && $OSTYPE != msys ]]; then
             mkdir -p $(dirname ${{ inputs.path }})
             ln -s $ITEM_PATH ${{ inputs.path }}
           else
diff --git a/.github/actions/save/action.yml b/.github/actions/save/action.yml
@@ -20,8 +20,18 @@ runs:
       if: ${{ inputs.enabled == 'true' }}
       shell: bash
       run: |
+        if [[ ! -d ${{ inputs.root }} ]]; then
+          echo "${{ inputs.root }} does not exist, cache disabled"
+          exit 0
+        fi
+
         TEMP_ITEM=$(mktemp -d -p ${{ inputs.root }})
-        cp -rT ${{ inputs.path }} $TEMP_ITEM
+        if [[ $OSTYPE == msys ]]; then
+          # dereference symlinks on Windows
+          cp -rLT ${{ inputs.path }} $TEMP_ITEM
+        else
+          cp -rT ${{ inputs.path }} $TEMP_ITEM
+        fi
         # ignore error if other job created a cache with the same key already
         if mv -T $TEMP_ITEM ${{ inputs.dest }}; then
           touch ${{ inputs.dest }}.hit
diff --git a/.github/workflows/e2e-windows.yml b/.github/workflows/e2e-windows.yml
@@ -4,6 +4,10 @@ run-name: ${{ inputs.run_name }}
 on:
   workflow_dispatch:
     inputs:
+      test_mode:
+        description: accuracy or performance
+        type: string
+        default: accuracy
       suite:
         description: Test suite
         type: choice
@@ -71,66 +75,22 @@ env:
   BENCHMARK_REPO: pytorch/benchmark
 
 jobs:
-  setup:
-    name: Setup
-    runs-on: linux
-    outputs:
-      suite: ${{ steps.set-matrix.outputs.suite }}
-      mode: ${{ steps.set-matrix.outputs.mode }}
-      dtype: ${{ steps.set-matrix.outputs.dtype }}
-      models: ${{ steps.set-matrix.outputs.models }}
-    timeout-minutes: 10
-    steps:
-      - name: Set matrix
-        id: set-matrix
-        run: |
-          if [[ -z "${{ inputs.suite }}" || "${{ inputs.suite }}" == "all" ]]; then
-            suite='["huggingface", "timm_models", "torchbench"]'
-          else
-            suite='["${{ inputs.suite }}"]'
-          fi
-          if [[ -z "${{ inputs.mode }}" || "${{ inputs.mode }}" == "all" ]]; then
-            mode='["inference", "inference-with-freezing", "training"]'
-          else
-            mode='["${{ inputs.mode }}"]'
-          fi
-          if [[ -z "${{ inputs.dtype }}" || "${{ inputs.dtype }}" == "all" ]]; then
-            dtype='["amp_bf16", "amp_fp16", "bfloat16", "float16", "float32"]'
-          else
-            dtype='["${{ inputs.dtype }}"]'
-          fi
-          if [[ -z "${{ inputs.models }}" ]]; then
-            models="subset"
-          else
-            models="${{ inputs.models }}"
-          fi
-          echo "suite=$suite" >> $GITHUB_OUTPUT
-          echo "mode=$mode" >> $GITHUB_OUTPUT
-          echo "dtype=$dtype" >> $GITHUB_OUTPUT
-          echo "models=$models" >> $GITHUB_OUTPUT
-
-      - name: Print inputs
-        run: |
-          cat <<EOF
-          ${{ toJSON(github.event.inputs) }}
-          EOF
-
-      - name: Print setup outputs
-        run: |
-          cat <<EOF
-          ${{ toJSON(steps.set-matrix.outputs) }}
-          EOF
-
   tests:
     name: Tests
-    needs: setup
     runs-on:
       - windows
       - ${{ inputs.runner_label }}
     timeout-minutes: 1440 # 24h
     strategy:
       fail-fast: false
     steps:
+      - name: Print inputs
+        shell: bash
+        run: |
+          cat <<EOF
+          ${{ toJSON(github.event.inputs) }}
+          EOF
+
       - name: Checkout repository
         uses: actions/checkout@v4
 
@@ -153,12 +113,6 @@ jobs:
         run:
           python -m venv .venv
 
-      # https://github.com/pytorch/data/blob/e316c5ca1ab2a4f69dd6d48e8fc9c6f8d0c7c468/README.md?plain=1#L6-L15
-      - name: Install pinned torchdata
-        run: |
-          .venv\Scripts\activate.ps1
-          pip install torchdata==0.9.0
-
       - name: Install PyTorch (source)
         run: |
           .venv\Scripts\activate.ps1
@@ -174,7 +128,7 @@ jobs:
         run: |
           .venv\Scripts\activate.ps1
           Invoke-BatchFile "C:\Program Files (x86)\Intel\oneAPI\setvars.bat"
-          python -c 'import torch;print(torch.__version__)' Tee-Object -Variable PYTORCH_VERSION
+          python -c 'import torch; print(torch.__version__)' | Tee-Object -Variable PYTORCH_VERSION
           echo "PYTORCH_VERSION=$PYTORCH_VERSION" | Out-File -FilePath $env:GITHUB_ENV -Encoding utf8 -Append
 
       - name: Clean up Triton cache
@@ -200,26 +154,88 @@ jobs:
           python -c 'import triton; print(triton.__version__)'
 
       - name: Identify pinned versions
+        shell: bash
         run: |
-          cd c:/pytorch
+          cd /c/pytorch
           echo "BENCHMARK_COMMIT_ID=$(<.github/ci_commit_pins/torchbench.txt)" | tee -a "$GITHUB_ENV"
           echo "TORCHVISION_COMMIT_ID=$(<.github/ci_commit_pins/vision.txt)" | tee -a "$GITHUB_ENV"
           echo "TORCHTEXT_COMMIT_ID=$(<.github/ci_commit_pins/text.txt)" | tee -a "$GITHUB_ENV"
           echo "TORCHAUDIO_COMMIT_ID=$(<.github/ci_commit_pins/audio.txt)" | tee -a "$GITHUB_ENV"
           echo "TRANSFORMERS_VERSION=$(<.ci/docker/ci_commit_pins/huggingface.txt)" | tee -a "$GITHUB_ENV"
           echo "TIMM_COMMIT_ID=$(<.ci/docker/ci_commit_pins/timm.txt)" | tee -a "$GITHUB_ENV"
 
+      - name: Install pinned torchdata
+        run: |
+          .venv\Scripts\activate.ps1
+          # Pinned torchdata
+          # https://github.com/pytorch/data/blob/e316c5ca1ab2a4f69dd6d48e8fc9c6f8d0c7c468/README.md?plain=1#L6-L15
+          pip install pyyaml pandas scipy numpy psutil pyre_extensions torchrec torchdata==0.9.0
+
+      - name: Install transformers package
+        if: inputs.suite == 'all' || inputs.suite == 'huggingface'
+        uses: ./.github/actions/install-dependency
+        with:
+          package: transformers
+          repository: huggingface/transformers
+          ref: ${{ env.TRANSFORMERS_VERSION }}
+          try-tag-prefix: v
+          extra-cache-key: ${{ env.PYTORCH_VERSION }}
+          workspace: /c/gh${{ github.run_id }}
+
+      - name: Install torchvision package
+        if: inputs.suite == 'all' || inputs.suite == 'timm_models' || inputs.suite == 'torchbench'
+        env:
+          DISTUTILS_USE_SDK: '1'
+        uses: ./.github/actions/install-dependency
+        with:
+          package: torchvision
+          repository: pytorch/vision
+          ref: ${{ env.TORCHVISION_COMMIT_ID }}
+          extra-cache-key: ${{ env.PYTORCH_VERSION }}
+          workspace: /c/gh${{ github.run_id }}
+
+      - name: Install torchtext package
+        if: inputs.suite == 'all' || inputs.suite == 'torchbench'
+        uses: ./.github/actions/install-dependency
+        with:
+          package: torchtext
+          repository: pytorch/text
+          ref: ${{ env.TORCHTEXT_COMMIT_ID }}
+          extra-cache-key: ${{ env.PYTORCH_VERSION }}
+          workspace: /c/gh${{ github.run_id }}
+
+      - name: Install torchaudio package
+        if: inputs.suite == 'all' || inputs.suite == 'torchbench'
+        uses: ./.github/actions/install-dependency
+        with:
+          package: torchaudio
+          repository: pytorch/audio
+          ref: ${{ env.TORCHAUDIO_COMMIT_ID }}
+          extra-cache-key: ${{ env.PYTORCH_VERSION }}
+          workspace: /c/gh${{ github.run_id }}
+
+      - name: Install timm package
+        if: inputs.suite == 'all' || inputs.suite == 'timm_models' || inputs.suite == 'torchbench'
+        uses: ./.github/actions/install-dependency
+        with:
+          package: timm
+          repository: huggingface/pytorch-image-models
+          ref: ${{ env.TIMM_COMMIT_ID }}
+          extra-cache-key: ${{ env.PYTORCH_VERSION }}
+          workspace: /c/gh${{ github.run_id }}
+
+      - name: Identify GPU
+        run: |
+          # Initializing oneAPI to enable sycl-ls, which is used in capture-hw-details.sh on Windows.
+          Invoke-BatchFile "C:\Program Files (x86)\Intel\oneAPI\setvars.bat"
+          bash -c './scripts/capture-hw-details.sh | tee -a $GITHUB_ENV'
+
       - name: Report environment details
         shell: bash
         run: |
-          mkdir -p ${{ env.NEW_WORKSPACE }}\inductor_log
+          mkdir -p /c/gh${{ github.run_id }}/inductor_log
+          cat <<EOF | tee /c/gh${{ github.run_id }}/inductor_log/e2e.env
           TIMESTAMP=$(date '+%Y%m%d%H%M%S')
-          echo "TIMESTAMP=$TIMESTAMP" >> "${GITHUB_ENV}"
-
-          source ./scripts/capture-hw-details.sh --quiet
-
-          cat <<EOF | tee ${{ env.NEW_WORKSPACE }}\inductor_log\.env
-          TIMESTAMP=$TIMESTAMP
           JOB_NAME=${{ join(inputs.*, '-') }}
           GITHUB_RUN_ID=$GITHUB_RUN_ID
           GITHUB_RUN_NUMBER=$GITHUB_RUN_NUMBER
diff --git a/scripts/install-pytorch.sh b/scripts/install-pytorch.sh
@@ -144,6 +144,7 @@ mkdir -p $BASE
 
 function pytorch_wheel_exists {
   if [[ ! -d $PYTORCH_PROJ/dist ]]; then
+    echo "check-wheel: $PYTORCH_PROJ/dist does not exist"
     return 1
   fi
   PYTHON_VERSION=$(python -c "import sys; print(f'{sys.version_info.major}{sys.version_info.minor}')")
@@ -158,10 +159,13 @@ function pytorch_wheel_exists {
   fi
   PYTORCH_WHEEL_NAME="torch-${PYTORCH_VERSION}+git${PYTORCH_COMMIT:0:7}-cp${PYTHON_VERSION}-cp${PYTHON_VERSION}-${PYTORCH_OS}_${PYTORCH_ARCH}.whl"
   if [[ -f $PYTORCH_PROJ/dist/$PYTORCH_WHEEL_NAME ]]; then
-    echo "**** $PYTORCH_WHEEL_NAME exists ****"
+    echo "check-wheel: $PYTORCH_WHEEL_NAME exists"
     return 0
   else
-    echo "**** $PYTORCH_WHEEL_NAME does not exist ****"
+    echo "check-wheel: $PYTORCH_WHEEL_NAME does not exist"
+    if [[ -d $PYTORCH_PROJ/dist ]]; then
+      echo "check-wheel: existing files:" $(cd $PYTORCH_PROJ/dist && ls)
+    fi
     return 1
   fi
 }
