While having access to the latest top secret Intel CPU and GPU hardware is one of the benefits of preparing science and engineering applications for the upcoming Aurora supercomputer, much of the open-source code for these applications is hosted in public repositories on GitHub, GitLab, Bitbucket, or similar services.

Adding Intel oneAPI compilers and toolkits to the CI pipelines for such projects can help catch many issues at the source as they are introduced during the development cycle. Fortunately this is easier than you might think.

This gist demonstrates how to add the Intel oneAPI DPC++ compiler and oneMKL library to GitHub actions for a project using CMake.

First, the oneAPI DPC++ compiler and oneMKL must be installed on the runner. For Ubuntu, this can be done using apt in a manner similar to an installation on your local machine. Notice that only the specific components needed are installed to avoid long build times.

name: CI

on:
  push:
    branches:
      - main
jobs:
  build:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v2

      - name: add oneAPI to apt
        shell: bash
        run: |
          cd /tmp
          wget https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB
          sudo apt-key add GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB
          rm GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB
          sudo add-apt-repository "deb https://apt.repos.intel.com/oneapi all main"
      
      - name: install oneAPI dpcpp compiler
        shell: bash
        run: |
          sudo apt update
          sudo apt install intel-oneapi-compiler-dpcpp-cpp
         
      - name: install oneAPI MKL library
        shell: bash
        run: |
          sudo apt install intel-oneapi-mkl-devel

Second, it is important to sourcesetvars.sh at the beginning of each step which uses the compilers or toolkits to ensure all necessary paths and environment variables are set.

      - name: configure
        shell: bash
        run: |
          source /opt/intel/oneapi/setvars.sh
          cmake -S . -B build \
          -DCMAKE_BUILD_TYPE="RelWithDebInfo" \
          -DCMAKE_INSTALL_PREFIX=install \
          -DCMAKE_CXX_COMPILER=dpcpp \
          -DCMAKE_C_COMPILER=icx \
          -DIntelDPCPP_DIR="/opt/intel/oneapi/compiler/latest/linux/cmake/SYCL" \
          -DMKL_ROOT="/opt/intel/oneapi/mkl/latest" \
          -DTBB_ROOT="/opt/intel/oneapi/tbb/latest"
      
      - name: build
        shell: bash
        run: |
          source /opt/intel/oneapi/setvars.sh
          cmake --build build

      - name: test
        shell: bash
        run: |
          source /opt/intel/oneapi/setvars.sh
          export SYCL_DEVICE_FILTER=opencl.cpu
          ctest --test-dir build --output-on-failure

Finally, the CMakeLists.txt file at the top of the source tree contains the following:

enable_language(CXX)
set(CMAKE_CXX_STANDARD 17)
set(CMAKE_POSITION_INDEPENDENT_CODE ON)

list(APPEND CMAKE_MODULE_PATH "${IntelDPCPP_DIR}")
find_package(IntelDPCPP REQUIRED)

find_package(MKL REQUIRED)

Notice the path to the DPC++ CMake package file must be added to CMAKE_MODULE_PATH.

Adding Intel oneAPI compilers and toolkits to the automated workflows for your open-source project is easier than you might think. Hopefully this helps with sharing your next great software project with the world.

Good luck and happy coding!