name: ios-simulator-gpu
on:
  push:
    branches: [master]
    paths:
    - '.github/workflows/ios-simulator-gpu.yml'
    - 'toolchains/ios.toolchain.cmake'
    - 'CMakeLists.txt'
    - 'cmake/**'
    - 'src/*'
    - 'src/layer/*'
    - 'src/layer/x86/**'
    - 'src/layer/arm/**'
  pull_request:
    branches: [master]
    paths:
    - '.github/workflows/ios-simulator-gpu.yml'
    - 'toolchains/ios.toolchain.cmake'
    - 'CMakeLists.txt'
    - 'cmake/**'
    - 'src/*'
    - 'src/layer/*'
    - 'src/layer/x86/**'
    - 'src/layer/arm/**'
concurrency:
  group: ios-simulator-gpu-${{ github.ref }}
  cancel-in-progress: true
env:
  DEVELOPER_DIR: /Applications/Xcode_13.4.1.app/Contents/Developer
  IOS_DEPLOYMENT_TARGET: '9.0'
  ENABLE_BITCODE: OFF
  ENABLE_ARC: OFF
  ENABLE_VISIBILITY: OFF
permissions:
  contents: read

jobs:
  ios-iphone-simulator-gpu:
    runs-on: macos-12
    steps:
    - uses: actions/checkout@v4
      with:
        submodules: true

    - name: cache-openmp
      id: cache-openmp
      uses: actions/cache@v3
      with:
        path: openmp-install
        key: openmp-ios-simulator-install-20230504
    - name: openmp
      if: steps.cache-openmp.outputs.cache-hit != 'true'
      run: |
        wget https://github.com/llvm/llvm-project/releases/download/llvmorg-11.0.0/openmp-11.0.0.src.tar.xz
        tar -xf openmp-11.0.0.src.tar.xz
        cd openmp-11.0.0.src
        sed -i'' -e '/.size __kmp_unnamed_critical_addr/d' runtime/src/z_Linux_asm.S
        sed -i'' -e 's/__kmp_unnamed_critical_addr/___kmp_unnamed_critical_addr/g' runtime/src/z_Linux_asm.S
    - name: openmp-build-i386
      if: steps.cache-openmp.outputs.cache-hit != 'true'
      run: |
        cd openmp-11.0.0.src
        mkdir -p build-i386 && cd build-i386
        cmake -DCMAKE_TOOLCHAIN_FILE=../../toolchains/ios.toolchain.cmake -DPLATFORM=SIMULATOR -DARCHS="i386" \
            -DDEPLOYMENT_TARGET=$IOS_DEPLOYMENT_TARGET -DENABLE_BITCODE=$ENABLE_BITCODE -DENABLE_ARC=$ENABLE_ARC -DENABLE_VISIBILITY=$ENABLE_VISIBILITY \
            -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=install \
            -DPERL_EXECUTABLE=/usr/local/bin/perl \
            -DLIBOMP_ENABLE_SHARED=OFF -DLIBOMP_OMPT_SUPPORT=OFF -DLIBOMP_USE_HWLOC=OFF ..
        cmake --build . -j 3
        cmake --build . --target install
    - name: openmp-build-x86_64
      if: steps.cache-openmp.outputs.cache-hit != 'true'
      run: |
        cd openmp-11.0.0.src
        mkdir -p build-x86_64 && cd build-x86_64
        cmake -DCMAKE_TOOLCHAIN_FILE=../../toolchains/ios.toolchain.cmake -DPLATFORM=SIMULATOR64 -DARCHS="x86_64" \
            -DDEPLOYMENT_TARGET=$IOS_DEPLOYMENT_TARGET -DENABLE_BITCODE=$ENABLE_BITCODE -DENABLE_ARC=$ENABLE_ARC -DENABLE_VISIBILITY=$ENABLE_VISIBILITY \
            -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=install \
            -DPERL_EXECUTABLE=/usr/local/bin/perl \
            -DLIBOMP_ENABLE_SHARED=OFF -DLIBOMP_OMPT_SUPPORT=OFF -DLIBOMP_USE_HWLOC=OFF ..
        cmake --build . -j 3
        cmake --build . --target install
    - name: openmp-build-arm64
      if: steps.cache-openmp.outputs.cache-hit != 'true'
      run: |
        cd openmp-11.0.0.src
        mkdir -p build-arm64 && cd build-arm64
        cmake -DCMAKE_TOOLCHAIN_FILE=../../toolchains/ios.toolchain.cmake -DPLATFORM=SIMULATORARM64 -DARCHS="arm64" \
            -DDEPLOYMENT_TARGET=$IOS_DEPLOYMENT_TARGET -DENABLE_BITCODE=$ENABLE_BITCODE -DENABLE_ARC=$ENABLE_ARC -DENABLE_VISIBILITY=$ENABLE_VISIBILITY \
            -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=install \
            -DPERL_EXECUTABLE=/usr/local/bin/perl \
            -DLIBOMP_ENABLE_SHARED=OFF -DLIBOMP_OMPT_SUPPORT=OFF -DLIBOMP_USE_HWLOC=OFF ..
        cmake --build . -j 3
        cmake --build . --target install
    - name: openmp-merge-fat-library
      if: steps.cache-openmp.outputs.cache-hit != 'true'
      run: |
        mkdir -p $GITHUB_WORKSPACE/openmp-install
        cp -a openmp-11.0.0.src/build-x86_64/install/include $GITHUB_WORKSPACE/openmp-install
        mkdir -p $GITHUB_WORKSPACE/openmp-install/lib
        lipo -create \
            openmp-11.0.0.src/build-i386/install/lib/libomp.a \
            openmp-11.0.0.src/build-x86_64/install/lib/libomp.a \
            openmp-11.0.0.src/build-arm64/install/lib/libomp.a \
            -o $GITHUB_WORKSPACE/openmp-install/lib/libomp.a

    - name: openmp-install
      run: |
        sudo cp $GITHUB_WORKSPACE/openmp-install/include/* $DEVELOPER_DIR/Platforms/iPhoneSimulator.platform/Developer/SDKs/iPhoneSimulator.sdk/usr/include
        sudo cp $GITHUB_WORKSPACE/openmp-install/lib/libomp.a $DEVELOPER_DIR/Platforms/iPhoneSimulator.platform/Developer/SDKs/iPhoneSimulator.sdk/usr/lib

    - name: vulkansdk
      run: |
        wget https://sdk.lunarg.com/sdk/download/1.3.236.0/mac/vulkansdk-macos-1.3.236.0.dmg?Human=true -O vulkansdk-macos-1.3.236.0.dmg
        hdiutil attach vulkansdk-macos-1.3.236.0.dmg
        sudo /Volumes/vulkansdk-macos-1.3.236.0/InstallVulkan.app/Contents/MacOS/InstallVulkan --root $GITHUB_WORKSPACE/vulkansdk-macos-1.3.236.0 --accept-licenses --default-answer --confirm-command install
        hdiutil detach /Volumes/vulkansdk-macos-1.3.236.0

    - name: build-x86_64
      run: |
        export VULKAN_SDK=$GITHUB_WORKSPACE/vulkansdk-macos-1.3.236.0/macOS
        mkdir build-x86_64 && cd build-x86_64
        cmake -DCMAKE_TOOLCHAIN_FILE=../toolchains/ios.toolchain.cmake -DPLATFORM=SIMULATOR64 -DARCHS="x86_64" \
            -DDEPLOYMENT_TARGET=$IOS_DEPLOYMENT_TARGET -DENABLE_BITCODE=$ENABLE_BITCODE -DENABLE_ARC=$ENABLE_ARC -DENABLE_VISIBILITY=$ENABLE_VISIBILITY \
            -DOpenMP_C_FLAGS="-Xclang -fopenmp" -DOpenMP_CXX_FLAGS="-Xclang -fopenmp" \
            -DOpenMP_C_LIB_NAMES="libomp" -DOpenMP_CXX_LIB_NAMES="libomp" \
            -DOpenMP_libomp_LIBRARY="$DEVELOPER_DIR/Platforms/iPhoneSimulator.platform/Developer/SDKs/iPhoneSimulator.sdk/usr/lib/libomp.a" \
            -DVulkan_INCLUDE_DIR=$GITHUB_WORKSPACE/vulkansdk-macos-1.3.236.0/MoltenVK/include \
            -DVulkan_LIBRARY=$GITHUB_WORKSPACE/vulkansdk-macos-1.3.236.0/MoltenVK/dylib/macOS/libMoltenVK.dylib \
            -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=install \
            -DNCNN_VULKAN=ON -DNCNN_BUILD_BENCHMARK=OFF ..
        cmake --build . -j 3
    - name: build-arm64
      run: |
        export VULKAN_SDK=$GITHUB_WORKSPACE/vulkansdk-macos-1.3.236.0/macOS
        mkdir build-arm64 && cd build-arm64
        cmake -DCMAKE_TOOLCHAIN_FILE=../toolchains/ios.toolchain.cmake -DPLATFORM=SIMULATORARM64 -DARCHS="arm64" \
            -DDEPLOYMENT_TARGET=$IOS_DEPLOYMENT_TARGET -DENABLE_BITCODE=$ENABLE_BITCODE -DENABLE_ARC=$ENABLE_ARC -DENABLE_VISIBILITY=$ENABLE_VISIBILITY \
            -DOpenMP_C_FLAGS="-Xclang -fopenmp" -DOpenMP_CXX_FLAGS="-Xclang -fopenmp" \
            -DOpenMP_C_LIB_NAMES="libomp" -DOpenMP_CXX_LIB_NAMES="libomp" \
            -DOpenMP_libomp_LIBRARY="$DEVELOPER_DIR/Platforms/iPhoneSimulator.platform/Developer/SDKs/iPhoneSimulator.sdk/usr/lib/libomp.a" \
            -DVulkan_INCLUDE_DIR=$GITHUB_WORKSPACE/vulkansdk-macos-1.3.236.0/MoltenVK/include \
            -DVulkan_LIBRARY=$GITHUB_WORKSPACE/vulkansdk-macos-1.3.236.0/MoltenVK/dylib/macOS/libMoltenVK.dylib \
            -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=install \
            -DNCNN_VULKAN=ON -DNCNN_BUILD_BENCHMARK=OFF ..
        cmake --build . -j 3