name: Nightly Test (AMD) on: schedule: - cron: '30 17 * * *' push: branches: - main paths: - "python/sglang/version.py" workflow_dispatch: inputs: aiter_ref: description: 'Override AITER commit (optional, leave empty to use Dockerfile default)' required: false type: string default: '' continue_on_error: description: 'Continue on error (do not fail the workflow on test failures)' required: false type: boolean default: true job_select: description: 'Select a job to run from dropdown (choose "all" to run all jobs)' required: false type: choice default: 'all' options: - 'all' - nightly-test-1-gpu-unit - nightly-accuracy-2-gpu - nightly-accuracy-2-gpu-vlm - nightly-perf-2-gpu-text - nightly-perf-2-gpu-vlm - nightly-4-gpu - nightly-accuracy-8-gpu - nightly-8-gpu-grok1-int4 - nightly-8-gpu-grok2 - nightly-8-gpu-deepseek-v31 - nightly-8-gpu-deepseek-v32 - nightly-8-gpu-deepseek-v32-mtp - nightly-8-gpu-deepseek-v3-kv-fp8 - nightly-8-gpu-kimi-k25 - nightly-8-gpu-qwen3-235b - nightly-8-gpu-qwen35 - nightly-8-gpu-glm5 - nightly-8-gpu-minimax-m25 - nightly-1-gpu-zimage-turbo - nightly-test-1-gpu-mi35x - nightly-accuracy-8-gpu-mi35x - nightly-8-gpu-mi35x-grok1-int4 - nightly-8-gpu-mi35x-grok2 - nightly-8-gpu-mi35x-deepseek-r1-mxfp4 - nightly-8-gpu-mi35x-deepseek-r1-mxfp4-kv-fp8 - nightly-8-gpu-mi35x-deepseek-r1-mxfp4-ar-fusion - nightly-accuracy-8-gpu-mi35x-deepseek-v32 - nightly-accuracy-8-gpu-mi35x-deepseek-v32-mtp - nightly-perf-8-gpu-mi35x-deepseek-v32-basic - nightly-perf-8-gpu-mi35x-deepseek-v32-mtp - nightly-8-gpu-mi35x-kimi-k25 - nightly-8-gpu-mi35x-qwen3-235b-mxfp4 - nightly-8-gpu-mi35x-qwen35 - nightly-8-gpu-mi35x-glm5 - nightly-8-gpu-mi35x-minimax-m25 job_filter: description: 'Or type comma-separated job names (overrides dropdown if non-empty)' required: false type: string default: '' workflow_call: inputs: ref: description: 'Git ref (branch, tag, or SHA) to test. If not provided, uses the default branch.' required: false type: string default: '' aiter_ref: description: 'Override AITER commit (optional, leave empty to use Dockerfile default)' required: false type: string default: '' job_filter: description: 'Select which job to run (leave empty or "all" to run all jobs)' required: false type: string default: 'all' continue_on_error: description: 'Continue on error (do not fail the workflow on test failures)' required: false type: boolean default: true env: AITER_COMMIT_OVERRIDE: ${{ inputs.aiter_ref }} concurrency: # When called via workflow_call with ref set, use a unique group per caller run to avoid # collisions with direct schedule/push triggers. We use inputs.ref (not github.event_name) # to detect this, because github.event_name inherits from the caller in workflow_call. group: nightly-test-amd-${{ inputs.ref && format('caller-{0}', github.run_id) || github.ref }} cancel-in-progress: ${{ !inputs.ref && github.event_name != 'workflow_call' }} jobs: # ============================================== MI30x Unit Tests ============================================== # 1-GPU Unit Tests - LoRA, debug utils, scheduler, etc. (MI30x only) nightly-test-1-gpu-unit: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-test-1-gpu-unit,')) runs-on: linux-mi325-1gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Nightly Unit Test (1-GPU) timeout-minutes: 90 run: | bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-1-gpu --nightly --timeout-per-file 900 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # ============================================== MI30x Accuracy Tests ============================================== # 2-GPU Accuracy Tests - GSM8K eval (MI30x only) nightly-accuracy-2-gpu: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-accuracy-2-gpu,')) runs-on: linux-mi325-2gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Nightly Test (2-GPU) run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd --nightly --timeout-per-file 7200 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # 2-GPU VLM Accuracy Tests - Vision-Language Models MMMU evaluation nightly-accuracy-2-gpu-vlm: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-accuracy-2-gpu-vlm,')) runs-on: linux-mi325-2gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Nightly Accuracy Test (2-GPU VLM MMMU) timeout-minutes: 180 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-2-gpu-vlm --nightly --timeout-per-file 7200 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # 2-GPU Text Models Performance Tests nightly-perf-2-gpu-text: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-perf-2-gpu-text,')) runs-on: linux-mi325-2gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Performance Test (2-GPU Text Models) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e SGLANG_USE_AITER=1 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-perf-text-2-gpu --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # 2-GPU VLM Performance Tests nightly-perf-2-gpu-vlm: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-perf-2-gpu-vlm,')) runs-on: linux-mi325-2gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Performance Test (2-GPU VLM Models) timeout-minutes: 180 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e SGLANG_USE_AITER=1 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-perf-vlm-2-gpu --nightly --timeout-per-file 7200 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # ============================================== MI30x 4-GPU Tests ============================================== # 4-GPU Nightly Tests - Dumper/Comparator E2E, VLM Encoder DP nightly-4-gpu: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-4-gpu,')) runs-on: linux-mi325-4gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Nightly Test (4-GPU) timeout-minutes: 120 run: | > github_summary.md bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-4-gpu --nightly --continue-on-error --timeout-per-file 3600 || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # 8-GPU Accuracy Tests - GPT-OSS, Grok1-FP8 (accuracy only) nightly-accuracy-8-gpu: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-accuracy-8-gpu,')) runs-on: linux-mi325-8gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Accuracy Test (8-GPU GPT-OSS) timeout-minutes: 180 run: | bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-gpt-oss --nightly --timeout-per-file 7200 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Accuracy Test (8-GPU Grok1-FP8) timeout-minutes: 60 run: | bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e RCCL_MSCCL_ENABLE=0 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-grok1-fp8 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # ============================================== MI30x Combined Accuracy + Performance Tests ============================================== # 8-GPU Grok1-INT4 (Accuracy + Performance combined) nightly-8-gpu-grok1-int4: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-grok1-int4,')) runs-on: linux-mi325-8gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Accuracy Test (8-GPU Grok1-INT4) timeout-minutes: 60 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e RCCL_MSCCL_ENABLE=0 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-grok1-int4 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Performance Test (8-GPU Grok1-INT4) timeout-minutes: 60 continue-on-error: true # Perf test failure doesn't fail the job if accuracy passed run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e RCCL_MSCCL_ENABLE=0 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-perf-8-gpu-grok1-int4 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # 8-GPU Grok2 (Accuracy + Performance combined) nightly-8-gpu-grok2: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-grok2,')) runs-on: linux-mi325-8gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Accuracy Test (8-GPU Grok2) timeout-minutes: 60 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e RCCL_MSCCL_ENABLE=0 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-grok2 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Performance Test (8-GPU Grok2) timeout-minutes: 60 continue-on-error: true # Perf test failure doesn't fail the job if accuracy passed run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e RCCL_MSCCL_ENABLE=0 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-perf-8-gpu-grok2 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # 8-GPU DeepSeek-V3.1 (Accuracy + Performance combined) nightly-8-gpu-deepseek-v31: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-deepseek-v31,')) runs-on: linux-mi325-8gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Accuracy Test (8-GPU DeepSeek-V3.1) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e SGLANG_USE_AITER=1 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-deepseek-v31 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Performance Test (8-GPU DeepSeek-V3.1) timeout-minutes: 300 continue-on-error: true # Perf test failure doesn't fail the job if accuracy passed run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e SGLANG_USE_ROCM700A=1 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-perf-8-gpu-deepseek-v31 --nightly --timeout-per-file 18000 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # 8-GPU DeepSeek-V3.2 (Basic Accuracy + Perf) nightly-8-gpu-deepseek-v32: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-deepseek-v32,')) runs-on: linux-mi325-8gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Accuracy Test (8-GPU DeepSeek-V3.2 Basic) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-deepseek-v32 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Performance Test (8-GPU DeepSeek-V3.2 Basic) timeout-minutes: 150 continue-on-error: true # Perf test failure doesn't fail the job if accuracy passed run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-perf-8-gpu-deepseek-v32-basic --nightly --timeout-per-file 5400 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # 8-GPU DeepSeek-V3.2 MTP (MTP Accuracy + Perf) nightly-8-gpu-deepseek-v32-mtp: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-deepseek-v32-mtp,')) runs-on: linux-mi325-8gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Accuracy Test (8-GPU DeepSeek-V3.2 MTP) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-deepseek-v32-mtp --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Performance Test (8-GPU DeepSeek-V3.2 MTP) timeout-minutes: 180 continue-on-error: true # Perf test failure doesn't fail the job if accuracy passed run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-perf-8-gpu-deepseek-v32-mtp --nightly --timeout-per-file 7200 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # 8-GPU DeepSeek-V3 KV FP8 (Basic + MTP with --kv-cache-dtype fp8_e4m3) nightly-8-gpu-deepseek-v3-kv-fp8: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-deepseek-v3-kv-fp8,')) runs-on: linux-mi325-8gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: DeepSeek-V3 KV FP8 Test (8-GPU Basic + MTP) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-8-gpu-deepseek-v3-kv-fp8 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # 8-GPU Kimi-K2.5 (Accuracy) nightly-8-gpu-kimi-k25: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-kimi-k25,')) runs-on: linux-mi325-8gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Accuracy Test (8-GPU Kimi-K2.5) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-kimi-k25 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} nightly-8-gpu-qwen3-235b: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-qwen3-235b,')) runs-on: linux-mi325-8gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Accuracy Test + Performance Test (8-GPU Qwen3) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-8-gpu-qwen3-235b --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # 8-GPU Qwen 3.5 (Accuracy) nightly-8-gpu-qwen35: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-qwen35,')) runs-on: linux-mi325-8gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh bash scripts/ci/amd/amd_ci_exec.sh pip install mistral-common "lm-eval[api]" - name: Accuracy Test (8-GPU Qwen 3.5) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-qwen35 --nightly --timeout-per-file 3600 || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} nightly-8-gpu-glm5: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-glm5,')) runs-on: linux-mi325-8gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh bash scripts/ci/amd/amd_ci_exec.sh pip install git+https://github.com/huggingface/transformers.git@96f807a33b75 - name: Accuracy Test (8-GPU GLM-5 NSA) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-glm5 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # 8-GPU MiniMax-M2.5 (Accuracy + Performance combined) nightly-8-gpu-minimax-m25: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-minimax-m25,')) runs-on: linux-mi325-8gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Accuracy Test (8-GPU MiniMax-M2.5) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e SGLANG_USE_AITER=1 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-minimax-m25 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Performance Test (8-GPU MiniMax-M2.5) timeout-minutes: 120 continue-on-error: true # Perf test failure doesn't fail the job if accuracy passed run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e SGLANG_USE_AITER=1 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-perf-8-gpu-minimax-m25 --nightly --timeout-per-file 5400 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # ============================================== MI30x Diffusion Tests ============================================== # 1-GPU Z-Image-Turbo (Diffusion T2I) nightly-1-gpu-zimage-turbo: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-1-gpu-zimage-turbo,')) runs-on: linux-mi325-1gpu-sglang steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: bash scripts/ci/amd/amd_ci_install_dependency.sh - name: Z-Image-Turbo Diffusion Test (1-GPU) timeout-minutes: 45 run: | bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ -e SGLANG_DIFFUSION_ARTIFACT_DIR="/sglang-checkout/diffusion-artifacts" \ pytest test/registered/amd/test_zimage_turbo.py -v -s ${{ inputs.continue_on_error && '|| true' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Upload generated images if: always() uses: actions/upload-artifact@v4 with: name: zimage-turbo-outputs path: diffusion-artifacts/ if-no-files-found: ignore retention-days: 30 # ============================================== MI35x Tests ============================================== # MI35x 1-GPU tests - platform-agnostic tests that may work on CDNA4 (gfx950) nightly-test-1-gpu-mi35x: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-test-1-gpu-mi35x,')) runs-on: linux-mi35x-gpu-1 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Nightly Test MI35x (1-GPU) timeout-minutes: 90 run: | bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-1-gpu-mi35x --nightly --timeout-per-file 900 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU Accuracy Tests - GPT-OSS (accuracy only) nightly-accuracy-8-gpu-mi35x: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-accuracy-8-gpu-mi35x,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Accuracy Test MI35x (8-GPU GPT-OSS) timeout-minutes: 180 run: | bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-8-gpu-mi35x --nightly --timeout-per-file 7200 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU Grok1-INT4 (Accuracy + Performance combined) nightly-8-gpu-mi35x-grok1-int4: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-mi35x-grok1-int4,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Accuracy Test MI35x (8-GPU Grok1-INT4) timeout-minutes: 90 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e RCCL_MSCCL_ENABLE=0 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-mi35x-grok1-int4 --nightly --timeout-per-file 5400 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Performance Test MI35x (8-GPU Grok1-INT4) timeout-minutes: 60 continue-on-error: true # Perf test failure doesn't fail the job if accuracy passed run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e RCCL_MSCCL_ENABLE=0 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-perf-8-gpu-mi35x-grok1-int4 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU Grok2 (Accuracy + Performance combined) nightly-8-gpu-mi35x-grok2: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-mi35x-grok2,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Accuracy Test MI35x (8-GPU Grok2) timeout-minutes: 60 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e RCCL_MSCCL_ENABLE=0 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-mi35x-grok2 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Performance Test MI35x (8-GPU Grok2) timeout-minutes: 60 continue-on-error: true # Perf test failure doesn't fail the job if accuracy passed run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e RCCL_MSCCL_ENABLE=0 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-perf-8-gpu-mi35x-grok2 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU DeepSeek-R1-MXFP4 (Accuracy + Performance combined) nightly-8-gpu-mi35x-deepseek-r1-mxfp4: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-mi35x-deepseek-r1-mxfp4,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Accuracy Test MI35x (8-GPU DeepSeek-R1-MXFP4) timeout-minutes: 180 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-8-gpu-mi35x-deepseek-r1-mxfp4 --nightly --timeout-per-file 7200 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Performance Test MI35x (8-GPU DeepSeek-R1-MXFP4) timeout-minutes: 300 continue-on-error: true # Perf test failure doesn't fail the job if accuracy passed run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 registered/amd/perf/mi35x/test_deepseek_r1_mxfp4_perf_mi35x.py || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU DeepSeek-R1-MXFP4 KV FP8 (Accuracy + Performance combined) nightly-8-gpu-mi35x-deepseek-r1-mxfp4-kv-fp8: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-mi35x-deepseek-r1-mxfp4-kv-fp8,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Accuracy Test MI35x (8-GPU DeepSeek-R1-MXFP4 KV FP8) timeout-minutes: 180 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-8-gpu-mi35x-deepseek-r1-mxfp4-kv-fp8 --nightly --timeout-per-file 7200 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Performance Test MI35x (8-GPU DeepSeek-R1-MXFP4 KV FP8) timeout-minutes: 300 continue-on-error: true # Perf test failure doesn't fail the job if accuracy passed run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 registered/amd/perf/mi35x/test_deepseek_r1_mxfp4_kv_fp8_perf_mi35x.py || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU DeepSeek-R1-MXFP4 AllReduce Fusion (Accuracy + Performance combined) nightly-8-gpu-mi35x-deepseek-r1-mxfp4-ar-fusion: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-mi35x-deepseek-r1-mxfp4-ar-fusion,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Accuracy Test MI35x (8-GPU DeepSeek-R1-MXFP4 AllReduce Fusion) timeout-minutes: 180 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-8-gpu-mi35x-deepseek-r1-mxfp4-ar-fusion --nightly --timeout-per-file 7200 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Performance Test MI35x (8-GPU DeepSeek-R1-MXFP4 AllReduce Fusion) timeout-minutes: 300 continue-on-error: true # Perf test failure doesn't fail the job if accuracy passed run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 registered/amd/perf/mi35x/test_deepseek_r1_mxfp4_ar_fusion_perf_mi35x.py || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU DeepSeek-V3.2 Accuracy Test nightly-accuracy-8-gpu-mi35x-deepseek-v32: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-accuracy-8-gpu-mi35x-deepseek-v32,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Accuracy Test MI35x (8-GPU DeepSeek-V3.2) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-8-gpu-mi35x-deepseek-v32 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU DeepSeek-V3.2 TP+MTP Accuracy Test nightly-accuracy-8-gpu-mi35x-deepseek-v32-mtp: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-accuracy-8-gpu-mi35x-deepseek-v32-mtp,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Accuracy Test MI35x (8-GPU DeepSeek-V3.2 TP+MTP) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-mi35x-deepseek-v32-mtp --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU DeepSeek-V3.2 Performance Test (Basic) nightly-perf-8-gpu-mi35x-deepseek-v32-basic: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-perf-8-gpu-mi35x-deepseek-v32-basic,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Performance Test MI35x (8-GPU DeepSeek-V3.2 Basic) timeout-minutes: 150 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-perf-8-gpu-mi35x-deepseek-v32-basic --nightly --timeout-per-file 5400 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU Kimi-K2.5 (Accuracy) nightly-8-gpu-mi35x-kimi-k25: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-mi35x-kimi-k25,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Accuracy Test MI35x (8-GPU Kimi-K2.5) timeout-minutes: 180 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-mi35x-kimi-k25 --nightly --timeout-per-file 7200 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU Qwen3-235B-MXFP4 (Accuracy + Performance) nightly-8-gpu-mi35x-qwen3-235b-mxfp4: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-mi35x-qwen3-235b-mxfp4,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Accuracy Test + Performance Test MI35x (8-GPU Qwen3-235B-MXFP4) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-8-gpu-mi35x-qwen3-235b-mxfp4 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU Qwen 3.5 (Accuracy) nightly-8-gpu-mi35x-qwen35: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-mi35x-qwen35,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate bash scripts/ci/amd/amd_ci_exec.sh pip install mistral-common "lm-eval[api]" - name: Accuracy Test MI35x (8-GPU Qwen 3.5) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-accuracy-8-gpu-mi35x-qwen35 --nightly --timeout-per-file 3600 || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} nightly-8-gpu-mi35x-glm5: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-mi35x-glm5,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate bash scripts/ci/amd/amd_ci_exec.sh pip install git+https://github.com/huggingface/transformers.git@96f807a33b75 - name: Accuracy Test MI35x (8-GPU GLM-5 NSA) timeout-minutes: 180 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-8-gpu-mi35x-glm5 --nightly --timeout-per-file 7200 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU MiniMax-M2.5 (Accuracy + Performance combined) nightly-8-gpu-mi35x-minimax-m25: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-8-gpu-mi35x-minimax-m25,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Accuracy Test MI35x (8-GPU MiniMax-M2.5) timeout-minutes: 120 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e SGLANG_USE_AITER=1 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-amd-8-gpu-mi35x-minimax-m25 --nightly --timeout-per-file 3600 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} - name: Performance Test MI35x (8-GPU MiniMax-M2.5) timeout-minutes: 120 continue-on-error: true # Perf test failure doesn't fail the job if accuracy passed run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e SGLANG_USE_AITER=1 \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-perf-8-gpu-mi35x-minimax-m25 --nightly --timeout-per-file 5400 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} # MI35x 8-GPU DeepSeek-V3.2 Performance Test (MTP) nightly-perf-8-gpu-mi35x-deepseek-v32-mtp: if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') && (!(inputs.job_filter || inputs.job_select) || (inputs.job_filter || inputs.job_select) == 'all' || contains(format(',{0},', inputs.job_filter || inputs.job_select), ',nightly-perf-8-gpu-mi35x-deepseek-v32-mtp,')) runs-on: linux-mi35x-gpu-8 steps: - name: Checkout code uses: actions/checkout@v4 with: ref: ${{ inputs.ref || github.ref }} - name: Setup docker run: | touch github_summary.md bash scripts/ci/amd/amd_ci_start_container.sh env: GITHUB_WORKSPACE: ${{ github.workspace }} - name: Install dependencies run: | bash scripts/ci/amd/amd_ci_install_dependency.sh # Install tabulate for run_suite.py (missing in MI35x container) bash scripts/ci/amd/amd_ci_exec.sh pip install tabulate - name: Performance Test MI35x (8-GPU DeepSeek-V3.2 MTP) timeout-minutes: 180 run: | > github_summary.md # Clear summary file bash scripts/ci/amd/amd_ci_exec.sh -w /sglang-checkout/test \ -e GITHUB_STEP_SUMMARY="/sglang-checkout/github_summary.md" \ python3 run_suite.py --hw amd --suite nightly-perf-8-gpu-mi35x-deepseek-v32-mtp --nightly --timeout-per-file 7200 ${{ inputs.continue_on_error && '--continue-on-error' || '' }} || TEST_EXIT_CODE=$? echo "$(> $GITHUB_STEP_SUMMARY || true exit ${TEST_EXIT_CODE:-0} check-all-jobs: if: always() && (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request' || github.event_name == 'workflow_dispatch') needs: # MI30x Unit Tests - nightly-test-1-gpu-unit # MI30x Accuracy Tests - nightly-accuracy-2-gpu - nightly-accuracy-2-gpu-vlm # MI30x 4-GPU Tests - nightly-4-gpu - nightly-accuracy-8-gpu # MI30x Performance Tests - excluded from check (perf failures don't block CI) # - nightly-perf-2-gpu-text # - nightly-perf-2-gpu-vlm # MI30x Combined Accuracy + Performance Tests - nightly-8-gpu-grok1-int4 - nightly-8-gpu-grok2 - nightly-8-gpu-deepseek-v31 - nightly-8-gpu-deepseek-v32 - nightly-8-gpu-deepseek-v32-mtp - nightly-8-gpu-deepseek-v3-kv-fp8 - nightly-8-gpu-kimi-k25 - nightly-8-gpu-qwen3-235b - nightly-8-gpu-qwen35 - nightly-8-gpu-glm5 - nightly-8-gpu-minimax-m25 # MI30x Diffusion Tests - nightly-1-gpu-zimage-turbo # MI35x jobs - nightly-test-1-gpu-mi35x - nightly-accuracy-8-gpu-mi35x - nightly-8-gpu-mi35x-grok1-int4 - nightly-8-gpu-mi35x-grok2 - nightly-8-gpu-mi35x-deepseek-r1-mxfp4 - nightly-8-gpu-mi35x-deepseek-r1-mxfp4-kv-fp8 - nightly-8-gpu-mi35x-deepseek-r1-mxfp4-ar-fusion - nightly-accuracy-8-gpu-mi35x-deepseek-v32 - nightly-accuracy-8-gpu-mi35x-deepseek-v32-mtp - nightly-8-gpu-mi35x-kimi-k25 - nightly-8-gpu-mi35x-qwen3-235b-mxfp4 - nightly-8-gpu-mi35x-qwen35 - nightly-8-gpu-mi35x-glm5 - nightly-8-gpu-mi35x-minimax-m25 # MI35x perf jobs excluded from check - perf failures don't block CI # - nightly-perf-8-gpu-mi35x-deepseek-v32-basic # - nightly-perf-8-gpu-mi35x-deepseek-v32-mtp runs-on: ubuntu-latest steps: - name: Check if any job failed run: | if [[ "${{ contains(needs.*.result, 'failure') }}" == "true" ]]; then echo "One or more nightly test jobs failed" exit 1 fi if [[ "${{ contains(needs.*.result, 'cancelled') }}" == "true" ]]; then echo "One or more nightly test jobs were cancelled" exit 1 fi echo "All nightly test jobs passed"