bencher

Bencher provides a continuous benchmarking workflow that runs the same benchmarks on identical bare‑metal hardware both locally and in CI. A CLI orchestrates execution, stores results, and the web console lets users query and graph performance over time. Customizable analytics detect regressions, and alerts can be posted to pull‑request checks to block merges that degrade speed.

The tool targets teams whose projects are sensitive to performance, including databases, compilers, browsers, runtimes, networking stacks, cryptographic libraries, and operating systems. By eliminating the variance typical of shared CI runners, Bencher reduces noise from over 30 % to under 2 %, making real performance changes observable.

Bencher integrates with existing benchmarking suites and can be used via a GitHub Action, allowing developers to catch performance bugs early in the development cycle and maintain reliable, reproducible metrics across code changes.