[TheRock CI] Adding MIOpen at HEAD (#2929)

* Adding MIOpen at HEAD

* Adding container and also adding CI run for .github paths

* Adding correct flags

* Adding patches

* Adding exception for ck

* rocm-libraries at new path

* adding global safe dir

* reorder

* Fixing paths

* Adding sharding

.github/workflows/therock-ci-linux.yml

@@ -27,30 +27,35 @@ jobs:
       TEATIME_FORCE_INTERACTIVE: 0
       AWS_SHARED_CREDENTIALS_FILE: /home/awsconfig/credentials.ini
     steps:
+      - name: "Checking out repository for rocm-libraries"
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          repository: "ROCm/rocm-libraries"
+
       - name: Checkout composable_kernel repository
         uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          path: "composable_kernel"
 
       - name: Checkout TheRock repository
         uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
         with:
           repository: "ROCm/TheRock"
-          ref: ec1c2ef4f2636bce7733fd8c95e1dbb6692c8a57
+          ref: 409f43ad9d564454bb1b23f8c8aa15d6b9d25200
           path: "TheRock"
 
       - name: Runner Health Settings
         run: |
-          df -h
-          cmake --version
-          echo "Installed Python versions:"
-          ls -d /opt/python
-          echo "python: $(which python), python3: $(which python3)"
-          echo "Git version: $(git --version)"
-          git config --global --add safe.directory $PWD
-          git config fetch.parallel 10
+          ./TheRock/build_tools/health_status.py
       
       - name: Fetch sources
         run: |
-          ./TheRock/build_tools/fetch_sources.py --jobs 12
+          ./TheRock/build_tools/fetch_sources.py --jobs 12 --no-include-rocm-libraries --no-include-ml-frameworks
+
+      - name: Patch rocm-libraries
+        run: |
+          git config --global --add safe.directory '*'
+          git -c user.name="therockbot" -c "user.email=therockbot@amd.com" am --whitespace=nowarn ./TheRock/patches/amd-mainline/rocm-libraries/*.patch
 
       - name: Install python deps
         run: |
@@ -92,32 +97,14 @@ jobs:
           aws-region: us-east-2
           role-to-assume: arn:aws:iam::692859939525:role/therock-artifacts-external
 
-      - name: Create Logs index Files and upload logs
+      - name: Post Build Upload
         if: always()
         run: |
-          python3 TheRock/build_tools/github_actions/create_log_index.py \
-            --build-dir=TheRock/build \
-            --amdgpu-family=${{ env.AMDGPU_FAMILIES }}
-
-          python3 TheRock/build_tools/github_actions/upload_build_logs_to_s3.py \
-            --build-dir=TheRock/build \
-            --run-id ${{ github.run_id }} \
-            --amdgpu-family ${{ env.AMDGPU_FAMILIES }}
-
-      - name: Upload artifacts
-        run: |
-          python TheRock/build_tools/github_actions/upload_build_artifacts.py \
+          python3 TheRock/build_tools/github_actions/post_build_upload.py \
             --run-id ${{ github.run_id }} \
             --amdgpu-family ${{ env.AMDGPU_FAMILIES }} \
-            --build-dir TheRock/build
-
-      - name: Add Links to Job Summary
-        if: always()
-        run: |
-          python TheRock/build_tools/github_actions/upload_build_summary.py \
-            --run-id ${{ github.run_id }} \
-            --amdgpu-family ${{ env.AMDGPU_FAMILIES }} \
-            --build-dir TheRock/build
+            --build-dir TheRock/build \
+            --upload
 
   therock-test-linux:
     name: "Test"

.github/workflows/therock-ci.yml

@@ -56,7 +56,14 @@ jobs:
     uses: ./.github/workflows/therock-ci-linux.yml
     secrets: inherit
     with:
-      cmake_options: "-DTHEROCK_ENABLE_COMPOSABLE_KERNEL=ON -DTHEROCK_ENABLE_MIOPEN=ON -DTHEROCK_ENABLE_ALL=OFF -DTHEROCK_USE_EXTERNAL_CK=ON -DTHEROCK_CK_SOURCE_DIR=../"
+      cmake_options: >-
+        -DTHEROCK_ENABLE_COMPOSABLE_KERNEL=ON 
+        -DTHEROCK_ENABLE_MIOPEN=ON 
+        -DTHEROCK_ENABLE_ALL=OFF 
+        -DTHEROCK_USE_EXTERNAL_COMPOSABLE_KERNEL=ON 
+        -DTHEROCK_COMPOSABLE_KERNEL_SOURCE_DIR=../composable_kernel
+        -DTHEROCK_USE_EXTERNAL_ROCM_LIBRARIES=ON
+        -DTHEROCK_ROCM_LIBRARIES_SOURCE_DIR=../
       amdgpu_families: "gfx94X-dcgpu"
       test_runs_on: "linux-mi325-1gpu-ossci-rocm"
 

.github/workflows/therock-test-component.yml

@@ -0,0 +1,71 @@
+name: Test component
+
+on:
+  workflow_call:
+    inputs:
+      artifact_run_id:
+        type: string
+        default: ""
+      amdgpu_families:
+        type: string
+      test_runs_on:
+        type: string
+      platform:
+        type: string
+      component:
+        type: string
+
+
+permissions:
+  contents: read
+
+jobs:
+  test_component:
+    name: 'Test ${{ fromJSON(inputs.component).job_name }} (shard ${{ matrix.shard }} of ${{ fromJSON(inputs.component).total_shards }})'
+    runs-on: ${{ inputs.test_runs_on }}
+    container:
+      image: ${{ inputs.platform == 'linux' && 'ghcr.io/rocm/no_rocm_image_ubuntu24_04@sha256:4150afe4759d14822f0e3f8930e1124f26e11f68b5c7b91ec9a02b20b1ebbb98' || null }}
+      options: --ipc host
+        --group-add video
+        --device /dev/kfd
+        --device /dev/dri
+        --group-add 992
+        --env-file /etc/podinfo/gha-gpu-isolation-settings
+    strategy:
+      fail-fast: false
+      matrix:
+        # The shard array is based on "total_shards" from "fetch_test_configurations.py"
+        # The test executable will shard based on the array. (ex: [1, 2, 3, 4] = four test shards)
+        shard: ${{ fromJSON(inputs.component).shard_arr }}
+    defaults:
+      run:
+        shell: bash
+    env:
+      VENV_DIR: ${{ github.workspace }}/.venv
+      ARTIFACT_RUN_ID: "${{ inputs.artifact_run_id != '' && inputs.artifact_run_id || github.run_id }}"
+      OUTPUT_ARTIFACTS_DIR: "./build"
+      THEROCK_BIN_DIR: "./build/bin"
+      AMDGPU_FAMILIES: ${{ inputs.amdgpu_families }}
+    steps:
+      - name: Checkout Repository
+        uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
+        with:
+          repository: "ROCm/TheRock"
+
+      - name: Run setup test environment workflow
+        uses: './.github/actions/setup_test_environment'
+        with:
+          ARTIFACT_RUN_ID: ${{ env.ARTIFACT_RUN_ID }}
+          AMDGPU_FAMILIES: ${{ inputs.amdgpu_families }}
+          OUTPUT_ARTIFACTS_DIR: ${{ env.OUTPUT_ARTIFACTS_DIR }}
+          VENV_DIR: ${{ env.VENV_DIR }}
+          FETCH_ARTIFACT_ARGS: ${{ fromJSON(inputs.component).fetch_artifact_args }}
+          IS_PR_FROM_FORK: ${{ github.event.pull_request.head.repo.fork }}
+
+      - name: Test
+        timeout-minutes: ${{ fromJSON(inputs.component).timeout_minutes }}
+        env:
+          SHARD_INDEX: ${{ matrix.shard }}
+          TOTAL_SHARDS: ${{ fromJSON(inputs.component).total_shards }}
+        run: |
+          ${{ fromJSON(inputs.component).test_script }}

.github/workflows/therock-test-packages.yml

@@ -37,41 +37,17 @@ jobs:
 
   test_components:
     name: 'Test ${{ matrix.components.job_name }}'
-    runs-on: ${{ inputs.test_runs_on }}
-    needs: configure_test_matrix
+    needs: [configure_test_matrix]
     # skip tests if no test matrix to run
     if: ${{ needs.configure_test_matrix.outputs.components != '[]' }}
     strategy:
       fail-fast: false
       matrix:
         components: ${{ fromJSON(needs.configure_test_matrix.outputs.components) }}
-    defaults:
-      run:
-        shell: bash
-    env:
-      VENV_DIR: ${{ github.workspace }}/.venv
-      ARTIFACT_RUN_ID: "${{ github.run_id }}"
-      OUTPUT_ARTIFACTS_DIR: ${{ github.workspace }}/build
-      THEROCK_BIN_DIR: "./build/bin"
-    steps:
-      - name: Checkout Repository
-        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
-        with:
-          repository: "ROCm/TheRock"
-
-      - name: Run setup test environment workflow
-        uses: './.github/actions/setup_test_environment'
-        with:
-          ARTIFACT_RUN_ID: ${{ env.ARTIFACT_RUN_ID }}
-          AMDGPU_FAMILIES: ${{ inputs.amdgpu_families }}
-          OUTPUT_ARTIFACTS_DIR: ${{ env.OUTPUT_ARTIFACTS_DIR }}
-          VENV_DIR: ${{ env.VENV_DIR }}
-          FETCH_ARTIFACT_ARGS: ${{ matrix.components.fetch_artifact_args }}
-          PLATFORM: ${{ inputs.platform }}
-          IS_PR_FROM_FORK: ${{ github.event.pull_request.head.repo.fork }}
-
-      - name: Test
-        timeout-minutes: ${{ matrix.components.timeout_minutes }}
-        run: |
-          if [ "${{ inputs.PLATFORM }}" == "linux" ]; then source ${VENV_DIR}/bin/activate ; else . ${VENV_DIR}/Scripts/activate ; fi
-          ${{ matrix.components.test_script }}
+    uses: './.github/workflows/therock-test-component.yml'
+    with:
+      artifact_run_id: ${{  github.run_id }}
+      amdgpu_families: ${{ inputs.amdgpu_families }}
+      test_runs_on: ${{ inputs.test_runs_on }}
+      platform: ${{ inputs.platform }}
+      component: ${{ toJSON(matrix.components) }}