hdf5/.github/workflows/nvhpc.yml

name: hdf5 dev nvhpc

# Triggers the workflow on a call from another workflow
on:
  workflow_call:
    inputs:
      build_mode:
        description: "release vs. debug build"
        required: true
        type: string

permissions:
  contents: read

env:
  CTEST_OUTPUT_ON_FAILURE: 1

jobs:
  nvhpc_build_and_test:
    name: "nvhpc ${{ inputs.build_mode }}"
    runs-on: ubuntu-latest
    env:
      CUDAVERDOT: "13.2"
      NVERDOT: "26.5"
      NVERDASH: "26-5"
      # CI stabilization: GitHub-hosted runners have no real InfiniBand, and the
      # Open MPI bundled with NVIDIA HPC-X sometimes has UCX select a
      # non-functional verbs interface (e.g. ud_verbs/mana_0), causing
      # intermittent "Failed to create UCP worker" / MPI_Init unreachable (-12)
      # failures. Force ob1 + tcp/shared-memory transports and skip UCX.
      OMPI_MCA_pml: ob1
      OMPI_MCA_btl: self,tcp,sm
      OMPI_MCA_mtl: ^ucx
      UCX_TLS: tcp,self,sm
      OMPI_MCA_btl_tcp_if_exclude: lo,docker0

    steps:
      - name: Get Sources
        uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1

      - name: Free Disk Space (Ubuntu)
        uses: jlumbroso/free-disk-space@54081f138730dfa15788a46383842cd2f914a1be # v1.3.1
        with:
          android: true
          dotnet: true
          haskell: true
          large-packages: true
          tool-cache: false
          swap-storage: true

      - name: Install Dependencies
        shell: bash
        run: |
          sudo apt-get update
          sudo apt-get install ninja-build graphviz
          sudo apt install libssl3 libssl-dev libcurl4 libcurl4-openssl-dev
          sudo apt install -y libaec-dev zlib1g-dev wget curl bzip2 flex bison cmake libzip-dev openssl build-essential

      - name: Install NVHPC
        shell: bash
        run: |
          curl https://developer.download.nvidia.com/hpc-sdk/ubuntu/DEB-GPG-KEY-NVIDIA-HPC-SDK | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-hpcsdk-archive-keyring.gpg
          echo 'deb [signed-by=/usr/share/keyrings/nvidia-hpcsdk-archive-keyring.gpg] https://developer.download.nvidia.com/hpc-sdk/ubuntu/amd64 /' | sudo tee /etc/apt/sources.list.d/nvhpc.list
          sudo apt-get update -y
          sudo apt-get install -y nvhpc-${{ env.NVERDASH }}
          echo "NVHPCSDK=/opt/nvidia/hpc_sdk" >> $GITHUB_ENV
          echo "OMPI_CXX=/opt/nvidia/hpc_sdk/Linux_x86_64/${{ env.NVERDOT }}/compilers/bin/nvc++" >> $GITHUB_ENV
          echo "OMPI_CC=/opt/nvidia/hpc_sdk/Linux_x86_64/${{ env.NVERDOT }}/compilers/bin/nvc" >> $GITHUB_ENV
          echo "OMPI_FC=/opt/nvidia/hpc_sdk/Linux_x86_64/${{ env.NVERDOT }}/compilers/bin/nvfortran" >> $GITHUB_ENV
          echo "CC=/opt/nvidia/hpc_sdk/Linux_x86_64/${{ env.NVERDOT }}/comm_libs/hpcx/bin/mpicc" >> $GITHUB_ENV
          echo "FC=/opt/nvidia/hpc_sdk/Linux_x86_64/${{ env.NVERDOT }}/comm_libs/hpcx/bin/mpifort" >> $GITHUB_ENV
          echo "LD_LIBRARY_PATH=/opt/nvidia/hpc_sdk/Linux_x86_64/${{ env.NVERDOT }}/cuda/${{ env.CUDAVERDOT }}/lib64:/opt/nvidia/hpc_sdk/Linux_x86_64/${{ env.NVERDOT }}/compilers/lib" >> $GITHUB_ENV
          echo "DESTDIR=/tmp" >> $GITHUB_ENV

      - name: Configure
        shell: bash
        run: |
          export PATH=/opt/nvidia/hpc_sdk/Linux_x86_64/${{ env.NVERDOT }}/comm_libs/hpcx/bin:/opt/nvidia/hpc_sdk/Linux_x86_64/${{ env.NVERDOT }}/compilers/bin:$PATH
          mkdir "${{ runner.workspace }}/build"
          cd "${{ runner.workspace }}/build"
          cmake -C $GITHUB_WORKSPACE/config/cmake/cacheinit.cmake -G Ninja \
          -DCMAKE_BUILD_TYPE=${{ inputs.build_mode }} \
          -DHDF5_ENABLE_PARALLEL:BOOL=ON \
          -DMPIEXEC_NUMPROC_FLAG:STRING=-np \
          -DMPIEXEC_MAX_NUMPROCS:STRING=2 \
          -DHDF5_BUILD_CPP_LIB:BOOL=OFF \
          -DHDF5_BUILD_FORTRAN:BOOL=ON \
          -DHDF5_BUILD_JAVA:BOOL=OFF \
          -DCMAKE_C_FLAGS="-tp=x86-64-v3" \
          -DCMAKE_Fortran_FLAGS="-tp=x86-64-v3" \
          -DCMAKE_CXX_FLAGS="-tp=x86-64-v3" \
          -DMPIEXEC_MAX_NUMPROCS:STRING="2" \
          -DMPIEXEC_PREFLAGS:STRING="--mca;opal_warn_on_missing_libcuda;0" \
          $GITHUB_WORKSPACE

      - name: Build
        shell: bash
        run: |
          cmake --build . --parallel 3 --config ${{ inputs.build_mode }}
        working-directory: ${{ runner.workspace }}/build

      # Skipping dt_arith and dtransform while we investigate long double failures
      - name: Run Tests
        shell: bash
        run: |
          ctest . -E "MPI_TEST|H5TEST-dt_arith" --parallel 2 -C ${{ inputs.build_mode }}
        working-directory: ${{ runner.workspace }}/build

      - name: Run Expected To Fail Tests
        shell: bash
        run: |
          ctest . -R "H5TEST-dt_arith" --parallel 2 -C ${{ inputs.build_mode }}
        working-directory: ${{ runner.workspace }}/build
        continue-on-error: true

      - name: Run Parallel Tests
        shell: bash
        run: |
          export UCX_WARN_UNUSED_ENV_VARS=n
          ctest . -R MPI_TEST -E "_by_chunk|_by_pattern" -C ${{ inputs.build_mode }}
        working-directory: ${{ runner.workspace }}/build