vllm/.buildkite/hardware_tests/amd.yaml

group: Hardware - AMD Build
steps:
  # Ensure ci_base is up-to-date before building the test image.
  # Compares a content hash of ci_base-affecting files against the remote
  # image label. If hashes match the build is skipped (< 30 s); if they
  # differ ci_base is rebuilt and pushed automatically.
  - label: "AMD: :docker: ensure ci_base"
    key: ensure-ci-base-amd
    depends_on: []
    device: amd_cpu
    no_plugin: true
    commands:
      - bash .buildkite/scripts/ci-bake-rocm.sh ci-base-rocm-ci-with-deps
    env:
      DOCKER_BUILDKIT: "1"
      VLLM_BAKE_FILE: "docker/docker-bake-rocm.hcl"
      PYTORCH_ROCM_ARCH: "gfx90a;gfx942;gfx950"
      REMOTE_VLLM: "1"
      VLLM_BRANCH: "$BUILDKITE_COMMIT"
    retry:
      automatic:
        - exit_status: -1  # Agent was lost
          limit: 1
        - exit_status: -10  # Agent was lost
          limit: 1

  - label: "AMD: :docker: build test image and artifacts"
    key: image-build-amd
    depends_on:
      - ensure-ci-base-amd
    device: amd_cpu
    no_plugin: true
    commands:
      - |
        if [[ "${ROCM_CI_ARTIFACT_ONLY:-0}" == "1" ]]; then
          echo "ROCM_CI_ARTIFACT_ONLY=1; building ROCm wheel artifact only"
          IMAGE_TAG="" bash .buildkite/scripts/ci-bake-rocm.sh test-rocm-ci-with-artifacts
        else
          bash .buildkite/scripts/ci-bake-rocm.sh test-rocm-ci-with-wheel
        fi
      - |
        docker run --rm --network=none --entrypoint /bin/bash "rocm/vllm-ci:${BUILDKITE_COMMIT}" -ec '
          if [ ! -d /vllm-workspace ]; then echo Missing directory: /vllm-workspace >&2; exit 1; fi
          if [ ! -d /vllm-workspace/tests ]; then echo Missing directory: /vllm-workspace/tests >&2; exit 1; fi
          if [ ! -d /vllm-workspace/src/vllm ]; then echo Missing directory: /vllm-workspace/src/vllm >&2; exit 1; fi
          if [ ! -x /vllm-workspace/src/vllm/vllm-rs ]; then echo Missing executable: /vllm-workspace/src/vllm/vllm-rs >&2; exit 1; fi
          command -v python3
          command -v uv
          command -v pytest
          if ! command -v amd-smi >/dev/null 2>&1 && ! command -v rocminfo >/dev/null 2>&1; then
            echo No ROCm CLI found in image >&2
            exit 1
          fi
          python3 - <<PY
        import torch, vllm
        print(torch.__version__)
        print(vllm.__version__)
        PY
          echo AMD image smoke OK
        '
    env:
      DOCKER_BUILDKIT: "1"
      VLLM_BAKE_FILE: "docker/docker-bake-rocm.hcl"
      PYTORCH_ROCM_ARCH: "gfx90a;gfx942;gfx950"
      IMAGE_TAG: "rocm/vllm-ci:$BUILDKITE_COMMIT"
      REMOTE_VLLM: "1"
      VLLM_BRANCH: "$BUILDKITE_COMMIT"
    retry:
      automatic:
        - exit_status: -1  # Agent was lost
          limit: 1
        - exit_status: -10  # Agent was lost
          limit: 1