.github/workflows/pkgci_regression_test.yml

# Copyright 2023 The IREE Authors
#
# Licensed under the Apache License v2.0 with LLVM Exceptions.
# See https://llvm.org/LICENSE.txt for license information.
# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception

name: PkgCI Regression Test
on:
  workflow_call:
    inputs:
      artifact_run_id:
        type: string
        default: ""
  workflow_dispatch:
    inputs:
      artifact_run_id:
        type: string
        default: ""

jobs:
  test_onnx:
    name: "test_onnx :: ${{ matrix.name }}"
    runs-on: ${{ matrix.runs-on }}
    strategy:
      fail-fast: false
      matrix:
        include:
          # CPU
          - name: cpu_llvm_sync
            config-file: onnx_cpu_llvm_sync.json
            numprocesses: auto
            runs-on: ubuntu-20.04

          # AMD GPU
          - name: amdgpu_rocm_rdna3
            numprocesses: 1
            config-file: onnx_gpu_rocm_rdna3.json
            runs-on: nodai-amdgpu-w7900-x86-64
          - name: amdgpu_vulkan
            numprocesses: 4
            config-file: onnx_gpu_vulkan.json
            runs-on: nodai-amdgpu-w7900-x86-64

          # NVIDIA GPU
          - name: nvidiagpu_cuda
            config-file: onnx_gpu_cuda.json
            numprocesses: 4
            runs-on:
              - self-hosted # must come first
              - runner-group=${{ github.event_name == 'pull_request' && 'presubmit' || 'postsubmit' }}
              - environment=prod
              - gpu # TODO(scotttodd): qualify further with vendor/model
              - os-family=Linux
          - name: nvidiagpu_vulkan
            config-file: onnx_gpu_vulkan.json
            numprocesses: 4
            runs-on:
              - self-hosted # must come first
              - runner-group=${{ github.event_name == 'pull_request' && 'presubmit' || 'postsubmit' }}
              - environment=prod
              - gpu # TODO(scotttodd): qualify further with vendor/model
              - os-family=Linux
    env:
      PACKAGE_DOWNLOAD_DIR: ${{ github.workspace }}/.packages
      IREERS_ARTIFACT_DIR: ${{ github.workspace }}/artifacts
      CONFIG_FILE_PATH: build_tools/pkgci/external_test_suite/${{ matrix.config-file }}
      NUMPROCESSES: ${{ matrix.numprocesses }}
      LOG_FILE_PATH: /tmp/iree_tests_onnx_${{ matrix.name }}_logs.json
      VENV_DIR: ${{ github.workspace }}/venv
    steps:
      - name: Checking out IREE repository
        uses: actions/checkout@v4.1.7
        with:
          submodules: false
      - uses: actions/setup-python@v5.1.0
        with:
          # Must match the subset of versions built in pkgci_build_packages.
          python-version: "3.11"
      - uses: actions/download-artifact@v4.1.7
        with:
          name: linux_x86_64_release_packages
          path: ${{ env.PACKAGE_DOWNLOAD_DIR }}
      - name: Setup venv
        run: |
          ./build_tools/pkgci/setup_venv.py ${VENV_DIR} \
            --artifact-path=${PACKAGE_DOWNLOAD_DIR} \
            --fetch-gh-workflow=${{ inputs.artifact_run_id }}

      - name: Check out external TestSuite repository
        uses: actions/checkout@v4.1.7
        with:
          repository: nod-ai/SHARK-TestSuite
          ref: 1da89a12a8b00dc77506d702f61300803b6240b7
          path: SHARK-TestSuite
          submodules: false
          lfs: false
      - name: Install external TestSuite Python requirements
        run: |
          source ${VENV_DIR}/bin/activate
          python -m pip install -r SHARK-TestSuite/iree_tests/requirements.txt
          pip install --no-compile --pre --upgrade -e SHARK-TestSuite/common_tools

      - name: Run external tests - ONNX test suite
        run: |
          source ${VENV_DIR}/bin/activate
          pytest SHARK-TestSuite/iree_tests/onnx/ \
              -rpfE \
              --numprocesses ${NUMPROCESSES} \
              --timeout=30 \
              --durations=20 \
              --no-skip-tests-missing-files \
              --config-files=${CONFIG_FILE_PATH} \
              --report-log=${LOG_FILE_PATH}
      - name: "Updating config file with latest XFAIL lists"
        if: failure()
        run: |
          source ${VENV_DIR}/bin/activate
          python SHARK-TestSuite/iree_tests/update_config_xfails.py \
            --log-file=${LOG_FILE_PATH} \
            --config-file=${CONFIG_FILE_PATH}
          cat ${CONFIG_FILE_PATH}
      - name: "Uploading new config file"
        if: failure()
        uses: actions/upload-artifact@v4.3.3
        with:
          name: ${{ matrix.config-file }}
          path: ${{ env.CONFIG_FILE_PATH }}

  test_models:
    name: "test_models :: ${{ matrix.name }}"
    runs-on: ${{ matrix.runs-on }}
    strategy:
      fail-fast: false

      # Note: these jobs should use persistent runners with local caches.
      # Downloading test files (50GB+) without a cache can take 20+ minutes.
      matrix:
        include:
          # CPU
          - name: cpu_llvm_task
            models-config-file: models_cpu_llvm_task.json
            iree-tests-cache: ~/iree_tests_cache
            runs-on: nodai-amdgpu-w7900-x86-64

          # AMD GPU
          - name: amdgpu_rocm_mi250_gfx90a
            models-config-file: models_gpu_rocm_gfx90a.json
            iree-tests-cache: /groups/aig_sharks/iree-tests-cache
            runs-on: nodai-amdgpu-mi250-x86-64
          - name: amdgpu_rocm_mi300_gfx942
            models-config-file: models_gpu_rocm_gfx942.json
            iree-tests-cache: ~/iree_tests_cache
            runs-on: nodai-amdgpu-mi300-x86-64
          - name: amdgpu_vulkan
            models-config-file: models_gpu_vulkan.json
            iree-tests-cache: ~/iree_tests_cache
            runs-on: nodai-amdgpu-w7900-x86-64

          # NVIDIA GPU
          # None at the moment. Could maybe use the persistent a100 runners:
          #   - self-hosted # must come first
          #   - runner-group=${{ needs.setup.outputs.runner-group }}
          #   - environment=${{ needs.setup.outputs.runner-env }}
          #   - a100
          #   - os-family=Linux
          # (note: would need to plumb the presubmit/postsubmit runner-group through to here too)
    env:
      PACKAGE_DOWNLOAD_DIR: ${{ github.workspace }}/.packages
      IREERS_ARTIFACT_DIR: ${{ github.workspace }}/artifacts
      IREE_TEST_FILES: ${{ matrix.iree-tests-cache }}
      IREE_TEST_PATH_EXTENSION: ${{ github.workspace }}/build_tools/pkgci/external_test_suite
      MODELS_CONFIG_FILE_PATH: build_tools/pkgci/external_test_suite/${{ matrix.models-config-file }}
      VENV_DIR: ${{ github.workspace }}/venv
      LD_LIBRARY_PATH: /home/esaimana/Python-3.11.9
    steps:
      # TODO(saienduri): Find alternative to this temporary step that manipulates permission of github actions
      # directory to be able to clean after every PR
      - name: Pre Checkout MI300 Step
        if: contains(matrix.name, 'gfx942')
        run: |
          sudo chmod -R 777 ~/actions-runner/_work
      - name: Checking out IREE repository
        uses: actions/checkout@v4.1.7
        with:
          submodules: false
      - uses: actions/setup-python@v5.1.0
        with:
          # Must match the subset of versions built in pkgci_build_packages.
          python-version: "3.11"
      - uses: actions/download-artifact@v4.1.7
        with:
          name: linux_x86_64_release_packages
          path: ${{ env.PACKAGE_DOWNLOAD_DIR }}
      - name: Setup venv
        run: |
          ./build_tools/pkgci/setup_venv.py ${VENV_DIR} \
            --artifact-path=${PACKAGE_DOWNLOAD_DIR} \
            --fetch-gh-workflow=${{ inputs.artifact_run_id }}

      # Out of tree tests
      - name: Check out external TestSuite repository
        uses: actions/checkout@v4.1.7
        with:
          repository: nod-ai/SHARK-TestSuite
          ref: 1da89a12a8b00dc77506d702f61300803b6240b7
          path: SHARK-TestSuite
          submodules: false
          lfs: true
      - name: Install external TestSuite Python requirements
        run: |
          source ${VENV_DIR}/bin/activate
          python3 -m pip install -r SHARK-TestSuite/iree_tests/requirements.txt
          pip install --no-compile --pre --upgrade -e SHARK-TestSuite/common_tools
      - name: Download remote files for real weight model tests
        run: |
          source ${VENV_DIR}/bin/activate
          python SHARK-TestSuite/iree_tests/download_remote_files.py --root-dir iree_tests/pytorch/models
          python SHARK-TestSuite/iree_tests/download_remote_files.py --root-dir iree_tests/sharktank

      - name: Run external tests - models with real weights
        if: "matrix.models-config-file != '' && !cancelled()"
        run: |
          source ${VENV_DIR}/bin/activate
          pytest \
            SHARK-TestSuite/iree_tests/pytorch/models \
            SHARK-TestSuite/iree_tests/sharktank \
            -rA \
            -k real_weights \
            --no-skip-tests-missing-files \
            --capture=no \
            --log-cli-level=info \
            --timeout=1200 \
            --durations=0 \
            --config-files=${MODELS_CONFIG_FILE_PATH}

  test_regression_suite:
    name: "test_regression_suite :: ${{ matrix.name }}"
    runs-on: ${{ matrix.runs-on }}
    strategy:
      fail-fast: false

      # Note: these jobs should use persistent runners with local caches.
      # Downloading test files (50GB+) without a cache can take 20+ minutes.
      matrix:
        include:
          # CPU
          - name: cpu_llvm_task
            models-config-file: models_cpu_llvm_task.json
            backend: cpu
            iree-tests-cache: ~/iree_tests_cache
            runs-on: nodai-amdgpu-w7900-x86-64

          # AMD GPU
          - name: amdgpu_rocm_mi250_gfx90a
            rocm-chip: gfx90a
            backend: rocm
            iree-tests-cache: /groups/aig_sharks/iree-tests-cache
            runs-on: nodai-amdgpu-mi250-x86-64
          - name: amdgpu_rocm_mi300_gfx942
            rocm-chip: gfx942
            backend: rocm
            iree-tests-cache: ~/iree_tests_cache
            runs-on: nodai-amdgpu-mi300-x86-64
    env:
      PACKAGE_DOWNLOAD_DIR: ${{ github.workspace }}/.packages
      IREERS_ARTIFACT_DIR: ${{ github.workspace }}/artifacts
      IREE_TEST_FILES: ${{ matrix.iree-tests-cache }}
      IREE_TEST_PATH_EXTENSION: ${{ github.workspace }}/build_tools/pkgci/external_test_suite
      VENV_DIR: ${{ github.workspace }}/venv
      LD_LIBRARY_PATH: /home/esaimana/Python-3.11.9
    steps:
      # TODO(saienduri): Find alternative to this temporary step that manipulates permission of github actions
      # directory to be able to clean after every PR
      - name: Pre Checkout MI300 Step
        if: contains(matrix.name, 'gfx942')
        run: |
          sudo chmod -R 777 ~/actions-runner/_work
      - name: Checking out IREE repository
        uses: actions/checkout@v4.1.7
        with:
          submodules: false
      - uses: actions/setup-python@v5.1.0
        with:
          # Must match the subset of versions built in pkgci_build_packages.
          python-version: "3.11"
      - uses: actions/download-artifact@v4.1.7
        with:
          name: linux_x86_64_release_packages
          path: ${{ env.PACKAGE_DOWNLOAD_DIR }}
      - name: Setup venv
        run: |
          ./build_tools/pkgci/setup_venv.py ${VENV_DIR} \
            --artifact-path=${PACKAGE_DOWNLOAD_DIR} \
            --fetch-gh-workflow=${{ inputs.artifact_run_id }}

      # TODO(#17344): regenerate .mlirbc files, test plat_rdna3_rocm on rocm
      # # In-tree tests
      # - name: Run experimental/regression_suite tests
      #   run: |
      #     source ${VENV_DIR}/bin/activate
      #     pytest \
      #       -rA -s -m "plat_host_cpu and presubmit" \
      #       experimental/regression_suite

      - name: "Running SDXL special model tests"
        if: "!cancelled()"
        run: |
          source ${VENV_DIR}/bin/activate
          pytest ./experimental/regression_suite/shark-test-suite-models/sdxl \
            -k ${{ matrix.backend }} \
            -rpfE \
            --capture=no \
            --log-cli-level=info \
            --timeout=1200 \
            --durations=0
        env:
          ROCM_CHIP: ${{ matrix.rocm-chip }}

      - name: "Running SD3 special model tests"
        if: "!cancelled()"
        run: |
          source ${VENV_DIR}/bin/activate
          pytest ./experimental/regression_suite/shark-test-suite-models/sd3 \
            -k ${{ matrix.backend }} \
            -rpfE \
            --capture=no \
            --log-cli-level=info \
            --timeout=1200 \
            --durations=0
        env:
          ROCM_CHIP: ${{ matrix.rocm-chip }}

      # Note: mi250 benchmark times are more lenient than mi300 (allowing about
      # 10% deviation from observed averages), since the mi250 runners we use
      # are more unstable and we care most about peak performance on mi300.
      - name: "Running SDXL rocm pipeline benchmark (mi250)"
        if: contains(matrix.name, 'rocm_mi250_gfx90a')
        run: |
          source ${VENV_DIR}/bin/activate
          pytest ./experimental/benchmarks/sdxl/benchmark_sdxl_rocm.py \
            --goldentime-rocm-e2e-ms 1450.0 \
            --goldentime-rocm-unet-ms 370.0 \
            --goldentime-rocm-clip-ms 18.5 \
            --goldentime-rocm-vae-ms 315.0 \
            --goldendispatch-rocm-unet 1714 \
            --goldendispatch-rocm-clip 1569 \
            --goldendispatch-rocm-vae 248 \
            --goldensize-rocm-unet-bytes 2280000  \
            --goldensize-rocm-clip-bytes 860000 \
            --goldensize-rocm-vae-bytes 840000 \
            --gpu-number 6 \
            --rocm-chip gfx90a \
            --log-cli-level=info \
            --retries 7
          echo "$(<job_summary.md )" >> $GITHUB_STEP_SUMMARY
          rm job_summary.md

      - name: "Running SDXL rocm pipeline benchmark (mi300)"
        if: contains(matrix.name, 'rocm_mi300_gfx942')
        run: |
          source ${VENV_DIR}/bin/activate
          pytest ./experimental/benchmarks/sdxl/benchmark_sdxl_rocm.py \
            --goldentime-rocm-e2e-ms 325.0 \
            --goldentime-rocm-unet-ms 77.0 \
            --goldentime-rocm-clip-ms 15.5 \
            --goldentime-rocm-vae-ms 74.0 \
            --goldendispatch-rocm-unet 1714 \
            --goldendispatch-rocm-clip 1569 \
            --goldendispatch-rocm-vae 248 \
            --goldensize-rocm-unet-bytes 2270000 \
            --goldensize-rocm-clip-bytes 860000  \
            --goldensize-rocm-vae-bytes 840000 \
            --gpu-number 0 \
            --rocm-chip gfx942 \
            --log-cli-level=info \
            --retries 7
          echo "$(<job_summary.md )" >> $GITHUB_STEP_SUMMARY