add hsa CI

nod-ai · Sep 3, 2024 · c1a15a9 · c1a15a9
1 parent 9041a72
commit c1a15a9
Show file tree

Hide file tree

Showing 2 changed files with 194 additions and 4 deletions.
diff --git a/.github/workflows/ci-hsa-linux.yml b/.github/workflows/ci-hsa-linux.yml
@@ -0,0 +1,104 @@
+name: CI HSA Linux
+
+on:
+  workflow_dispatch:
+    inputs:
+      force_debug_with_tmate:
+        type: boolean
+        description: 'Run the build with tmate session'
+        required: false
+        default: false
+      debug_with_tmate:
+        type: boolean
+        description: 'Run the build with a tmate session ONLY in case of failure'
+        required: false
+        default: false
+      release:
+        type: boolean
+        description: 'Release to latest'
+        required: false
+        default: false
+  pull_request:
+  merge_group:
+  push:
+    branches:
+      - amd-staging
+
+concurrency:
+  group: ci-build-test-cpp-hsa-linux-${{ github.event.number || github.sha }}
+  cancel-in-progress: true
+
+jobs:
+  build_hsa:
+    name: Build HSA (linux)
+    runs-on: nod-ai-shared-cpubuilder-manylinux-x86_64
+    steps:
+      - name: Configure local git mirrors
+        run: |
+          /gitmirror/scripts/trigger_update_mirrors.sh
+          /gitmirror/scripts/git_config.sh
+
+      - name: "Checking out repository"
+        env:
+          BRANCH_NAME: ${{ github.ref }}
+          REPO_ADDRESS: ${{ github.server_url }}/${{ github.repository }}
+        run: |
+          git init
+          git remote add origin $REPO_ADDRESS
+          git -c protocol.version=2 fetch --depth 1 origin $BRANCH_NAME
+          git reset --hard FETCH_HEAD
+          git -c submodule."third_party/torch-mlir".update=none -c submodule."third_party/stablehlo".update=none -c submodule."src/runtime_src/core/common/aiebu".update=none submodule update --init --recursive --depth 1 --single-branch -j 10
+
+      - name: Install deps
+        run: |
+          dnf install -y almalinux-release-devel
+          yum install -y elfutils-libelf-devel p7zip p7zip-plugins \
+                         sudo ncurses-compat-libs openssh vim-common
+
+      - name: Build and install libnuma
+        working-directory: ${{ github.workspace }}/third_party/ROCR-Runtime
+        run: |
+          curl --silent -L \
+            https://github.com/numactl/numactl/releases/download/v2.0.18/numactl-2.0.18.tar.gz \
+            -o numactl-2.0.18.tar.gz
+          tar -xf numactl-2.0.18.tar.gz
+          pushd numactl-2.0.18
+          ./configure
+          # i have no idea why this is necessary
+          # but without it you get something about "can't cd into dir"
+          sed -i '7563s/`cd "$dir" && pwd`/$dir/g' libtool
+          make install
+          popd
+
+      - name: Hack ROCR
+        working-directory: ${{ github.workspace }}/third_party/ROCR-Runtime
+        run: |
+          sed -i 's/amdgcn-amd-amdhsa/amdgcn-amd-amdhsa -nogpulib/g' runtime/hsa-runtime/core/runtime/blit_shaders/CMakeLists.txt
+          sed -i 's/amdgcn-amd-amdhsa/amdgcn-amd-amdhsa -nogpulib/g' runtime/hsa-runtime/core/runtime/trap_handler/CMakeLists.txt
+          sed -i 's/amdgcn-amd-amdhsa/amdgcn-amd-amdhsa -nogpulib/g' runtime/hsa-runtime/image/blit_src/CMakeLists.txt
+
+      - name: Get Clang
+        working-directory: ${{ github.workspace }}/third_party/ROCR-Runtime
+        run: |
+          pip download mlir==20.0.0.2024090301+amdgpu.b6597f52 -f https://makslevental.github.io/wheels
+          unzip -q mlir-*.whl
+
+      - name: Build ROCR distro
+        working-directory: ${{ github.workspace }}/third_party/ROCR-Runtime
+        run: |
+          rocr_dir="$PWD"
+          build_rocr_dir="$PWD/rocr-build"
+          mkdir -p "$build_rocr_dir"
+          build_rocr_dir="$(cd $build_rocr_dir && pwd)"
+          rocr_install_dir="$PWD/rocr-install"
+          
+          cmake -GNinja \
+            -DCMAKE_BUILD_TYPE=Release \
+            -DCMAKE_INSTALL_PREFIX="$rocr_install_dir" \
+            -DClang_DIR=$PWD/mlir/lib/cmake/clang \
+            -DLLVM_DIR=$PWD/mlir/lib/cmake/mlir \
+            -DIMAGE_SUPPORT=OFF \
+            -S "$rocr_dir" -B "$build_rocr_dir"
+          
+          cmake --build "$build_rocr_dir" --target install
+          tar -cf rocr-${GITHUB_SHA::8}.tar rocr-install
diff --git a/.github/workflows/ci-linux.yml b/.github/workflows/ci-linux.yml
@@ -17,6 +17,88 @@ concurrency:
   cancel-in-progress: true
 
 jobs:
+  build_hsa:
+    name: Build HSA (linux)
+    runs-on: nod-ai-shared-cpubuilder-manylinux-x86_64
+    steps:
+      - name: Configure local git mirrors
+        run: |
+          /gitmirror/scripts/trigger_update_mirrors.sh
+          /gitmirror/scripts/git_config.sh
+
+      - name: "Checking out repository"
+        env:
+          BRANCH_NAME: ${{ github.ref }}
+          REPO_ADDRESS: ${{ github.server_url }}/${{ github.repository }}
+        run: |
+          git init
+          git remote add origin $REPO_ADDRESS
+          git -c protocol.version=2 fetch --depth 1 origin $BRANCH_NAME
+          git reset --hard FETCH_HEAD
+          git -c submodule."third_party/torch-mlir".update=none -c submodule."third_party/stablehlo".update=none -c submodule."src/runtime_src/core/common/aiebu".update=none submodule update --init --recursive --depth 1 --single-branch -j 10
+
+      - name: Install deps
+        run: |
+          dnf install -y almalinux-release-devel
+          yum install -y elfutils-libelf-devel p7zip p7zip-plugins \
+                         sudo ncurses-compat-libs openssh vim-common
+
+      - name: Build and install libnuma
+        working-directory: ${{ github.workspace }}/third_party/ROCR-Runtime
+        run: |
+          curl --silent -L \
+            https://github.com/numactl/numactl/releases/download/v2.0.18/numactl-2.0.18.tar.gz \
+            -o numactl-2.0.18.tar.gz
+          tar -xf numactl-2.0.18.tar.gz
+          pushd numactl-2.0.18
+          ./configure
+          # i have no idea why this is necessary
+          # but without it you get something about "can't cd into dir"
+          sed -i '7563s/`cd "$dir" && pwd`/$dir/g' libtool
+          make install
+          popd
+
+      - name: Hack ROCR
+        working-directory: ${{ github.workspace }}/third_party/ROCR-Runtime
+        run: |
+          sed -i 's/amdgcn-amd-amdhsa/amdgcn-amd-amdhsa -nogpulib/g' runtime/hsa-runtime/core/runtime/blit_shaders/CMakeLists.txt
+          sed -i 's/amdgcn-amd-amdhsa/amdgcn-amd-amdhsa -nogpulib/g' runtime/hsa-runtime/core/runtime/trap_handler/CMakeLists.txt
+          sed -i 's/amdgcn-amd-amdhsa/amdgcn-amd-amdhsa -nogpulib/g' runtime/hsa-runtime/image/blit_src/CMakeLists.txt
+
+      - name: Get compatible Clang
+        working-directory: ${{ github.workspace }}/third_party/ROCR-Runtime
+        run: |
+          pip download mlir==20.0.0.2024090301+amdgpu.b6597f52 -f https://makslevental.github.io/wheels
+          unzip -q mlir-*.whl
+
+      - name: Build ROCR distro
+        working-directory: ${{ github.workspace }}/third_party/ROCR-Runtime
+        run: |
+          rocr_dir="$PWD"
+          build_rocr_dir="$PWD/rocr-build"
+          mkdir -p "$build_rocr_dir"
+          build_rocr_dir="$(cd $build_rocr_dir && pwd)"
+          rocr_install_dir="$PWD/rocr-install"
+          
+          cmake -GNinja \
+            -DCMAKE_BUILD_TYPE=Release \
+            -DCMAKE_INSTALL_PREFIX="$rocr_install_dir" \
+            -DClang_DIR=$PWD/mlir/lib/cmake/clang \
+            -DLLVM_DIR=$PWD/mlir/lib/cmake/mlir \
+            -DIMAGE_SUPPORT=OFF \
+            -S "$rocr_dir" -B "$build_rocr_dir"
+          
+          cmake --build "$build_rocr_dir" --target install
+          tar -cf rocr-${GITHUB_SHA::8}.tar rocr-install
+
+      - name: Upload artifacts
+        uses: actions/upload-artifact@v4
+        if: ${{ !cancelled() }}
+        with:
+          name: linux_hsa_x86_64_release_packages
+          path: ${{ github.workspace }}/third_party/ROCR-Runtime/rocr-*.tar
+          if-no-files-found: error
+
   build_and_ctest:
     name: Build and Test (linux, ASSERTIONS)
     runs-on: nod-ai-shared-cpubuilder-manylinux-x86_64
@@ -55,7 +137,7 @@ jobs:
         run: |
           dnf install -y almalinux-release-devel epel-release
           yum remove -y openssl-devel zlib-devel || true
-          yum install -y protobuf-devel protobuf-compiler libnuma-devel tmate
+          yum install -y protobuf-devel protobuf-compiler tmate
 
       - name: Python deps
         run: |
@@ -68,10 +150,14 @@ jobs:
           key:  ${{ env.CACHE_KEY }}
           restore-keys: linux-build-test-cpp-
 
-      - name: Build ROCT/ROCR
+      - name: Download artifacts
+        uses: actions/download-artifact@v4
+        with:
+          name: linux_hsa_x86_64_release_packages
+
+      - name: Extract artifact
         run: |
-          export cache_dir="${{ env.CACHE_DIR }}"
-          bash build_tools/ci/build_roct_rocr.sh
+          tar -xvf rocr-*.tar
           echo "hsa-runtime64_ROOT=$PWD/rocr-install" >> $GITHUB_ENV
 
       - name: Build packages