bench.yml

name: 'Benchmark'

on:
  pull_request:
  pull_request_review:
    types: [submitted]
  workflow_dispatch:

concurrency:
  group: ${{ github.workflow }}-${{ github.ref }}${{ github.event_name == 'pull_request_review' && format('-review-{0}', github.run_id) || '' }}
  cancel-in-progress: true

jobs:
  file-changes:
    name: Detect File Changes
    if: >
      github.event_name != 'pull_request_review' ||
      github.event.review.user.type != 'Bot'
    runs-on: 'ubuntu-latest'
    outputs:
      checkall: ${{ steps.changes.outputs.checkall }}
    steps:
      - name: Clone
        uses: actions/checkout@v4

      - name: Detect Changes
        uses: dorny/paths-filter@v3
        id: changes
        with:
          filters: ".github/file-filter.yml"

  self:
    name: "${{ matrix.name }} (${{ matrix.device }}${{ matrix.interface != 'none' && format('-{0}', matrix.interface) || '' }})"
    if: ${{ github.repository=='MFlowCode/MFC' && needs.file-changes.outputs.checkall=='true' && github.event.pull_request.draft != true && ((github.event_name=='pull_request_review' && github.event.review.state=='approved') || (github.event_name=='pull_request' && (github.event.pull_request.user.login=='sbryngelson' || github.event.pull_request.user.login=='wilfonba')) || github.event_name=='workflow_dispatch') }}
    needs: file-changes
    strategy:
      fail-fast: false
      matrix:
        include:
          - cluster: phoenix
            name: Georgia Tech | Phoenix (NVHPC)
            group: phoenix
            labels: gt
            flag: p
            device: cpu
            interface: none
            build_script: ""
          - cluster: phoenix
            name: Georgia Tech | Phoenix (NVHPC)
            group: phoenix
            labels: gt
            flag: p
            device: gpu
            interface: acc
            build_script: ""
          - cluster: phoenix
            name: Georgia Tech | Phoenix (NVHPC)
            group: phoenix
            labels: gt
            flag: p
            device: gpu
            interface: omp
            build_script: ""
          - cluster: frontier
            name: Oak Ridge | Frontier (CCE)
            group: phoenix
            labels: frontier
            flag: f
            device: gpu
            interface: acc
            build_script: "bash .github/workflows/frontier/build.sh gpu acc"
          - cluster: frontier
            name: Oak Ridge | Frontier (CCE)
            group: phoenix
            labels: frontier
            flag: f
            device: gpu
            interface: omp
            build_script: "bash .github/workflows/frontier/build.sh gpu omp"
          - cluster: frontier_amd
            name: Oak Ridge | Frontier (AMD)
            group: phoenix
            labels: frontier
            flag: famd
            device: gpu
            interface: omp
            build_script: "bash .github/workflows/frontier_amd/build.sh gpu omp"
    continue-on-error: ${{ matrix.cluster == 'frontier' || matrix.cluster == 'frontier_amd' }}
    runs-on:
      group: ${{ matrix.group }}
      labels: ${{ matrix.labels }}
    timeout-minutes: 480
    steps:
      - name: Clean stale output files
        run: rm -f *.out

      - name: Clone - PR
        uses: actions/checkout@v4
        with:
          path: pr
          clean: false

      - name: Clone - Master
        uses: actions/checkout@v4
        with:
          repository: MFlowCode/MFC
          ref: master
          path: master
          clean: false

      - name: Fetch Dependencies
        if: matrix.build_script != ''
        timeout-minutes: 150
        run: |
          (cd pr     && ${{ matrix.build_script }}) &
          pid1=$!
          (cd master && ${{ matrix.build_script }}) &
          pid2=$!
          e1=0; e2=0
          wait $pid1 || e1=$?
          wait $pid2 || e2=$?
          if [ $e1 -ne 0 ] || [ $e2 -ne 0 ]; then
            echo "Build failures: pr=$e1 master=$e2"
            exit 1
          fi

      - name: Bench (Master v. PR)
        run: bash pr/.github/scripts/run_parallel_benchmarks.sh ${{ matrix.device }} ${{ matrix.interface }} ${{ matrix.cluster }}

      - name: Cancel SLURM Jobs
        if: cancelled()
        run: |
          find . -name "*.slurm_job_id" | while read -r f; do
            job_id=$(cat "$f")
            echo "Cancelling SLURM job $job_id"
            scancel "$job_id" 2>/dev/null || true
          done

      - name: Generate & Post Comment
        if: always()
        run: |
          (cd pr && . ./mfc.sh load -c ${{ matrix.flag }} -m g)
          (cd pr && ./mfc.sh bench_diff ../master/bench-${{ matrix.device }}-${{ matrix.interface }}.yaml ../pr/bench-${{ matrix.device }}-${{ matrix.interface }}.yaml)

      - name: Print Logs
        if:   always()
        run: |
          cat pr/bench-${{ matrix.device }}-${{ matrix.interface }}.* 2>/dev/null || true
          cat master/bench-${{ matrix.device }}-${{ matrix.interface }}.* 2>/dev/null || true

      - name: Print Per-Case Logs
        if: always()
        run: |
          passed=() failed=()
          for out in pr/build/benchmarks/*/*.out master/build/benchmarks/*/*.out; do
            [ -f "$out" ] || continue
            [ -f "${out%.out}.yaml" ] && passed+=("$out") || failed+=("$out")
          done

          echo "=== Per-Case Summary: ${#failed[@]} failed, ${#passed[@]} passed ==="
          for out in "${failed[@]}"; do echo "  [FAILED] $out"; done
          for out in "${passed[@]}"; do echo "  [PASSED] $out"; done

          if [ ${#failed[@]} -gt 0 ]; then
            echo ""
            echo "=== Failed Case Logs ==="
            for out in "${failed[@]}"; do
              echo "--- $out ---"
              cat "$out"
              echo ""
            done
          fi

      # All other runners (non-Phoenix) just run without special env
      - name: Archive Logs (Frontier)
        if: always() && matrix.cluster != 'phoenix'
        uses: actions/upload-artifact@v4
        with:
          name: ${{ matrix.cluster }}-${{ matrix.device }}-${{ matrix.interface }}
          path: |
            pr/bench-${{ matrix.device }}-${{ matrix.interface }}.*
            pr/build/benchmarks/*
            master/bench-${{ matrix.device }}-${{ matrix.interface }}.*
            master/build/benchmarks/*