cuda.yml

name: 🐧 CUDA

on: [push, pull_request]

concurrency:
  group: ${{ github.ref }}-${{ github.head_ref }}-cuda
  cancel-in-progress: true

jobs:
# Ref.:
#   https://gitlab.com/nvidia/container-images/cuda/-/blob/master/dist/ubuntu18.04/10.1/base/Dockerfile
#   https://github.com/ComputationalRadiationPhysics/picongpu/blob/0.5.0/share/picongpu/dockerfiles/ubuntu-1604/Dockerfile
#   https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/
  build_nvcc:
    name: NVCC 11.0.2 SP
    runs-on: ubuntu-18.04
    if: github.event.pull_request.draft == false
    env:
      CXXFLAGS: "-Werror"
      CMAKE_GENERATOR: Ninja
    steps:
    - uses: actions/checkout@v2
    - name: install dependencies
      run: |
        .github/workflows/dependencies/nvcc11.sh
        export CEI_SUDO="sudo"
        cmake-easyinstall --prefix=/usr/local \
          git+https://github.com/openPMD/openPMD-api.git@0.14.3 \
          -DopenPMD_USE_PYTHON=OFF    \
          -DBUILD_TESTING=OFF         \
          -DBUILD_EXAMPLES=OFF        \
          -DBUILD_CLI_TOOLS=OFF       \
          -DCMAKE_VERBOSE_MAKEFILE=ON
    - name: build WarpX
      run: |
        export PATH=/usr/local/nvidia/bin:/usr/local/cuda/bin:${PATH}
        export LD_LIBRARY_PATH=/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda/lib64:${LD_LIBRARY_PATH}
        which nvcc || echo "nvcc not in PATH!"

        cmake -S . -B build_sp         \
          -DCMAKE_VERBOSE_MAKEFILE=ON  \
          -DWarpX_COMPUTE=CUDA         \
          -DWarpX_EB=ON                \
          -DWarpX_LIB=ON               \
          -DAMReX_CUDA_ARCH=6.0        \
          -DWarpX_OPENPMD=ON           \
          -DWarpX_openpmd_internal=OFF \
          -DWarpX_PRECISION=SINGLE     \
          -DWarpX_PSATD=ON             \
          -DAMReX_CUDA_ERROR_CROSS_EXECUTION_SPACE_CALL=ON \
          -DAMReX_CUDA_ERROR_CAPTURE_THIS=ON
        cmake --build build_sp -j 2

        python3 -m pip install --upgrade pip setuptools wheel
        export WARPX_MPI=ON
        PYWARPX_LIB_DIR=$PWD/build_sp/lib python3 -m pip wheel .
        python3 -m pip install *.whl

  build_nvhpc21-9-nvcc:
    name: NVHPC@21.9 NVCC/NVC++ Release [tests]
    runs-on: ubuntu-20.04
    if: github.event.pull_request.draft == false
    # For NVHPC, Ninja is slower than the default:
    #env:
    #  CMAKE_GENERATOR: Ninja
    steps:
    - uses: actions/checkout@v2
    - name: Dependencies
      run: .github/workflows/dependencies/nvhpc.sh
    - name: Build & Install
      run: |
        source /etc/profile.d/modules.sh
        module load /opt/nvidia/hpc_sdk/modulefiles/nvhpc/21.9
        which nvcc || echo "nvcc not in PATH!"
        which nvc++ || echo "nvc++ not in PATH!"
        which nvc || echo "nvc not in PATH!"
        nvcc --version
        nvc++ --version
        nvc --version
        cmake --version

        export CC=$(which nvc)
        export CXX=$(which nvc++)
        export CUDACXX=$(which nvcc)
        export CUDAHOSTCXX=${CXX}

        cmake -S . -B build            \
          -DCMAKE_VERBOSE_MAKEFILE=ON  \
          -DWarpX_COMPUTE=CUDA         \
          -DWarpX_EB=ON                \
          -DWarpX_LIB=ON               \
          -DAMReX_CUDA_ARCH=8.0        \
          -DWarpX_OPENPMD=ON           \
          -DWarpX_PSATD=ON             \
          -DAMReX_CUDA_ERROR_CROSS_EXECUTION_SPACE_CALL=ON \
          -DAMReX_CUDA_ERROR_CAPTURE_THIS=ON
        cmake --build build -j 2

        # work-around for mpi4py 3.1.1 build system issue with using
        # a GNU-built Python executable with non-GNU Python modules
        #   https://github.com/mpi4py/mpi4py/issues/114
        export CFLAGS="-noswitcherror"

        python3 -m pip install --upgrade pip setuptools wheel
        export WARPX_MPI=ON
        PYWARPX_LIB_DIR=$PWD/build/lib python3 -m pip wheel .
        python3 -m pip install *.whl