hipdnn-superbuild-ci.yml

# Copyright (c) Advanced Micro Devices, Inc., or its affiliates.
# SPDX-License-Identifier: MIT
#
# Lightweight CI for hipDNN + providers using the rocm-libraries superbuild
# and pre-built ROCm wheels (no TheRock source build required).

name: hipDNN Superbuild CI

on:
  pull_request:
    types: [opened, synchronize, reopened]
    paths:
      - 'projects/hipdnn/**'
      - 'dnn-providers/**'

permissions:
  contents: read

concurrency:
  group: ${{ github.workflow }}-${{ github.event.number || github.sha }}
  cancel-in-progress: true

jobs:
  hipdnn-windows-superbuild:
    name: HipDNN Windows Superbuild
    runs-on: azure-windows-scale-rocm
    defaults:
      run:
        shell: bash
    env:
      BUILD_DIR: B:\build
      VENV_PATH: B:\rocm_wheels
      GPU_TARGET: gfx1151
    steps:
      - name: Runner health status
        shell: pwsh
        run: |
          Write-Host "=== OS ==="
          (Get-CimInstance Win32_OperatingSystem).Caption
          Write-Host "=== CPU ==="
          $cpu = Get-CimInstance Win32_Processor
          Write-Host "$($cpu.Name)  ($($cpu.NumberOfLogicalProcessors) logical cores)"
          Write-Host "=== Memory ==="
          $os = Get-CimInstance Win32_OperatingSystem
          $totalGB = [math]::Round($os.TotalVisibleMemorySize / 1MB, 1)
          $freeGB  = [math]::Round($os.FreePhysicalMemory / 1MB, 1)
          Write-Host "Total: ${totalGB} GB  Free: ${freeGB} GB"
          Write-Host "=== Disk ==="
          Get-PSDrive -PSProvider FileSystem |
            Select-Object Name,
              @{N='Used(GB)';E={[math]::Round($_.Used/1GB,1)}},
              @{N='Free(GB)';E={[math]::Round($_.Free/1GB,1)}},
              Root |
            Format-Table -AutoSize | Out-String | Write-Host

      - name: Configure git longpaths
        run: git config --global core.longpaths true

      - name: Checkout repository
        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
        with:
          sparse-checkout: |
            .dvc
            .github
            cmake
            dnn-providers
            projects/hipdnn
            projects/miopen
            test

      - name: Set up Python
        uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
        with:
          python-version: '3.12'

      - name: Configure MSVC
        uses: ilammy/msvc-dev-cmd@0b201ec74fa43914dc39ae48a89fd1d8cb592756 # v1.13.0

      - name: Install build tools
        run: |
          choco source disable -n=chocolatey
          choco source add -n=internal -s http://10.0.167.96:8081/repository/choco-group/ --priority=1
          choco install --no-progress -y ccache
          choco install --no-progress -y ninja --version 1.12.1

      - name: Install ROCm wheels
        shell: pwsh
        run: |
          Write-Host "Creating Python virtual environment at $env:VENV_PATH"
          python -m venv $env:VENV_PATH
          & "$env:VENV_PATH\Scripts\Activate.ps1"

          Write-Host "Installing ROCm wheels from nightlies (target: $env:GPU_TARGET)"
          pip install --index-url "https://rocm.nightlies.amd.com/v2/$env:GPU_TARGET/" "rocm[libraries,devel]"
          if ($LASTEXITCODE -ne 0) { throw "Failed to install ROCm wheels" }

          Write-Host "Initializing ROCm SDK"
          rocm-sdk init
          if ($LASTEXITCODE -ne 0) { throw "Failed to initialize ROCm SDK" }

          $RocmSdkPath = "$env:VENV_PATH\Lib\site-packages\_rocm_sdk_devel" -replace '\\', '/'
          Write-Host "ROCM_SDK_PATH=$RocmSdkPath"
          echo "ROCM_SDK_PATH=$RocmSdkPath" >> $env:GITHUB_ENV

      - name: Configure superbuild
        run: |
          cmake --preset hipdnn-providers-all -GNinja \
            -DROCM_PATH="$ROCM_SDK_PATH" \
            -DCMAKE_PROGRAM_PATH="$ROCM_SDK_PATH/lib/llvm/bin" \
            -DGPU_TARGETS="$GPU_TARGET" \
            -DROCM_LIBS_ENABLE_ROOT_CTEST=ON \
            -DENABLE_CLANG_FORMAT=OFF \
            -DENABLE_CLANG_TIDY=OFF

      - name: Build
        run: cmake --build build

      # Windows tests have a few issues where we are loading plugins and thats causing issues with super build tests
      # as they end up trying to init a gpu on a non gpu machine.  This work is ticketed to be fixed later.
      #- name: Add ROCm bin to PATH
      #  run: echo "$ROCM_SDK_PATH/bin" >> "$GITHUB_PATH"

      #- name: Run tests
      #  run: ctest --test-dir build --output-on-failure

  hipdnn-linux-superbuild:
    name: HipDNN Linux Superbuild
    runs-on: azure-linux-scale-rocm
    env:
      GPU_TARGET: gfx94X-dcgpu
    steps:
      - name: Runner health status
        run: |
          echo "=== OS ==="
          cat /etc/os-release | head -2
          echo "=== CPU ==="
          lscpu | grep -E 'Model name|^CPU\(s\)'
          echo "=== Memory ==="
          free -h | grep Mem
          echo "=== Disk ==="
          df -h /

      - name: Checkout repository
        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
        with:
          sparse-checkout: |
            .dvc
            .github
            cmake
            dnn-providers
            projects/hipdnn
            projects/miopen
            test

      - name: Set environment
        run: echo "VENV_PATH=${{ runner.temp }}/rocm_wheels" >> "$GITHUB_ENV"

      - name: Set up Python
        uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
        with:
          python-version: '3.12'

      - name: Install build tools
        run: |
          sudo apt-get update
          sudo apt-get install -y ninja-build ccache lsb-release wget software-properties-common gnupg
          wget https://apt.llvm.org/llvm.sh
          chmod +x llvm.sh
          sudo ./llvm.sh 20
          sudo apt-get install -y clang-tidy-20 clang-tools-20

      - name: Install ROCm wheels
        run: |
          echo "Creating Python virtual environment at $VENV_PATH"
          python -m venv "$VENV_PATH"
          source "$VENV_PATH/bin/activate"

          echo "Installing ROCm wheels from nightlies (target: $GPU_TARGET)"
          pip install --index-url "https://rocm.nightlies.amd.com/v2/$GPU_TARGET/" "rocm[libraries,devel]"

          echo "Initializing ROCm SDK"
          rocm-sdk init

          SITE_PACKAGES=$("$VENV_PATH/bin/python" -c "import site; print(site.getsitepackages()[0])")
          ROCM_SDK_PATH="$SITE_PACKAGES/_rocm_sdk_devel"
          echo "ROCM_SDK_PATH=$ROCM_SDK_PATH"
          echo "ROCM_SDK_PATH=$ROCM_SDK_PATH" >> "$GITHUB_ENV"

      - name: Configure superbuild
        run: |
          cmake --preset hipdnn-providers-all -GNinja \
            -DROCM_PATH="$ROCM_SDK_PATH" \
            -DCMAKE_PROGRAM_PATH="$ROCM_SDK_PATH/lib/llvm/bin" \
            -DGPU_TARGETS="$GPU_TARGET" \
            -DROCM_LIBS_ENABLE_ROOT_CTEST=ON \
            -DENABLE_CLANG_FORMAT=OFF \
            -DENABLE_CLANG_TIDY=ON \
            -DCMAKE_CXX_CLANG_TIDY=/usr/bin/clang-tidy-20

      - name: Build
        run: cmake --build build

      - name: Run tests
        run: ctest --test-dir build --output-on-failure

  hipdnn_build_summary:
    name: hipDNN Superbuild Summary
    if: always()
    needs:
      - hipdnn-windows-superbuild
      - hipdnn-linux-superbuild
    runs-on: ubuntu-24.04
    steps:
      - name: Report results
        run: |
          echo '${{ toJson(needs) }}'
          FAILED_JOBS="$(echo '${{ toJson(needs) }}' \
            | jq --raw-output \
            'map_values(select(.result!="success" and .result!="skipped")) | keys | join(",")' \
          )"
          if [[ "${FAILED_JOBS}" != "" ]]; then
            echo "Failed jobs: ${FAILED_JOBS}"
            exit 1
          fi