agentflow/.github/workflows/benchmark-arm.yml at main · brownjuly2003-code/agentflow · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
name: ARM Benchmark

# Dispatch-only benchmark on real ARM server hardware.
# Runs on the GitHub-hosted arm64 runner for public repositories
# (ubuntu-24.04-arm: Cobalt 100, Arm Neoverse N2, Armv9-A, 4 vCPU) — the
# available ARM server class for this project. Not a required check; uploaded
# `.artifacts/` files are runtime artifacts. Tracked evidence is created only
# by separate review/promotion.

on:
  workflow_dispatch:

permissions:
  contents: read

jobs:
  benchmark-arm:
    runs-on: ubuntu-24.04-arm
    timeout-minutes: 30
    steps:
      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
      - uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
        with:
          python-version: "3.11"
      - name: Install dependencies
        run: bash scripts/ci_sync.sh perf
      - name: Record host metadata
        run: |
          mkdir -p .artifacts/benchmark
          {
            echo "# ARM benchmark host metadata"
            echo
            echo "Generated by workflow run ${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}"
            echo
            echo '```'
            uname -a
            echo "runner-label: ubuntu-24.04-arm"
            nproc
            lscpu
            free -h
            echo '```'
          } > .artifacts/benchmark/arm-host-metadata.md
      - name: Run benchmark
        # DuckDB-profile benchmark: no ClickHouse service on the ARM runner,
        # and recorded evidence compares against DuckDB-era numbers.
        env:
          SERVING_BACKEND: duckdb
        run: >
          python scripts/run_benchmark.py
          --report-path .artifacts/benchmark/arm-benchmark.md
          --results-json .artifacts/benchmark/arm-current.json
      - name: Upload benchmark artifacts
        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
        with:
          name: arm-benchmark
          path: |
            .artifacts/benchmark/arm-host-metadata.md
            .artifacts/benchmark/arm-benchmark.md
            .artifacts/benchmark/arm-current.json
          if-no-files-found: error