Benchmarks workflow (learning-at-home/hivemind)

The Benchmarks workflow from learning-at-home/hivemind, explained and optimized by Latchkey.

What it does

This is the Benchmarks workflow from the learning-at-home/hivemind repository, a real project running GitHub Actions. It is shown here with attribution under its MIT license.

Below, Latchkey shows a faster, safer version produced by its optimization engine.

The workflow

name: Benchmarks

on:
  push:
    branches: [ master ]
  pull_request:

concurrency:
  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
  cancel-in-progress: true

jobs:
  run_benchmarks:

    runs-on: ubuntu-latest
    timeout-minutes: 10
    steps:
      - uses: actions/checkout@v4
      - name: Set up Python
        uses: actions/setup-python@v5
        with:
          python-version: 3.11
      - name: Cache dependencies
        uses: actions/cache@v4
        with:
          path: ~/.cache/pip
          key: Key-v1-3.11-${{ hashFiles('requirements.txt') }}-${{ hashFiles('requirements-dev.txt') }}
      - name: Install dependencies
        run: |
          python -m pip install --upgrade pip
          pip install -r requirements.txt
          pip install -r requirements-dev.txt
      - name: Build bitsandbytes
        run: |
          pip install bitsandbytes==0.45.2
      - name: Build hivemind
        run: |
          pip install .
      - name: Benchmark
        run: |
          cd benchmarks
          python benchmark_throughput.py --preset minimalistic
          python benchmark_tensor_compression.py
          python benchmark_dht.py

The same workflow, on Latchkey

Estimated ~20% faster on cache hits, plus fewer wasted runs and a safer supply chain. Added and changed lines are highlighted.

name: Benchmarks
 
on:
  push:
    branches: [ master ]
  pull_request:
 
concurrency:
  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
  cancel-in-progress: true
 
jobs:
  run_benchmarks:
 
    runs-on: latchkey-small
    timeout-minutes: 10
    steps:
      - uses: actions/checkout@v4
      - name: Set up Python
        uses: actions/setup-python@v5
        with:
          cache: 'pip'
          python-version: 3.11
      - name: Cache dependencies
        uses: actions/cache@v4
        with:
          path: ~/.cache/pip
          key: Key-v1-3.11-${{ hashFiles('requirements.txt') }}-${{ hashFiles('requirements-dev.txt') }}
      - name: Install dependencies
        run: |
          python -m pip install --upgrade pip
          pip install -r requirements.txt
          pip install -r requirements-dev.txt
      - name: Build bitsandbytes
        run: |
          pip install bitsandbytes==0.45.2
      - name: Build hivemind
        run: |
          pip install .
      - name: Benchmark
        run: |
          cd benchmarks
          python benchmark_throughput.py --preset minimalistic
          python benchmark_tensor_compression.py
          python benchmark_dht.py
 

What changed

• Run on Latchkey managed runners with one line (runs-on), which apply the fixes below automatically and self-heal transient failures. This example uses latchkey-small; pick the runner size that fits the job.
• Cache dependency installs on the setup step so they are served from cache.

What Latchkey heals here

This workflow has steps that commonly fail on transient issues (network, registries, flaky browsers). On Latchkey managed runners they are detected, retried, and self-healed instead of failing your build:

• Dependency installs

This workflow runs 1 job per trigger. On Latchkey the same minutes cost up to 58% less than GitHub-hosted, with zero queue time.

Actions used in this workflow