CI Bench v1:GitHub Actions runners 的開放基準測試
CI Bench v1: an open benchmark of GitHub Actions runners
CI Bench v1 公開 GitHub Actions runners 基準測試,涵蓋熱門開源項目的 12 個 CI 工作負載;測試方法、工作流程、分析及原始工作記錄均已公開。測試的九種配置中,Vitko Runners 配合 warm starts 和測試並行化速度最快、每項工作的託管成本最低;RunsOn 使用 spot instances 在極高用量下每項工作成本較低,但耗時約兩倍。
CI job times and costs are an increasingly large bottleneck for coding agents. Today we're releasing CI Bench v1, our open benchmark of Github Actions runners on 12 CI workloads for popular open-source projects.
The method, workflows, analysis and every raw job record are public in our GitHub repo, measured October 7 to 11, 2026.
Among the nine configurations measured, Vitko Runners with warm starts and test parallelization was the fastest and had the lowest managed cost per job; RunsOn on spot instances costs less per job at very high volume, while taking about twice as long.
Full disclosure: we sell Vitko Runners, now in private beta.
來源:DEV Community · dev.to
