GitHub Pages benchmark scoreboard

Publish full machine benchmarks with detailed system specs.

torture-bench runs on the user's machine, records platform details, module-level results, verdicts, notes, and raw JSON, then publishes everything into this static scoreboard.

GitHub Pages is static, so it cannot run the benchmark in the browser. The benchmark runs locally via script, then commits detailed output into docs/data/ for Pages to display.
Latest raw JSON Repository

Linux / macOS

One command or repo-local helper

bash tools/run_benchmark_publish.sh --duration 10 --tune --label "My Linux box"

# Alternative from a local clone
./run_and_submit.sh --publish-pages

Windows PowerShell

One-liner or repo-local helper

$env:BENCH_LABEL="My Windows box"
$env:BENCH_PUBLISH_PAGES="1"
irm https://raw.githubusercontent.com/sunprojectca/Benchmarks/main/bench.ps1 | iex

REM Alternative from a local clone
run_and_submit.bat --publish-pages

Submission notes

User selections

  • Choose duration, thread count, tune mode, included/excluded modules, label, and notes.
  • The repo-local helpers stay local by default and only touch docs/data/ when you pass --publish-pages.
  • With --publish-pages, the helper scripts commit the generated Pages data and try to push it.
  • Raw benchmark JSON is copied into docs/data/runs/ for download.
  • The scoreboard stores full platform specs, config, verdict, warnings, and per-module detail.

Runs

0

Detailed records currently published

Best composite

0.00

Highest score among visible runs

Fair verdicts

0

Runs without acceleration warnings

Platforms

0

Distinct OS / architecture combinations

Explore the scoreboard

Filter by platform, verdict, or module

Showing all runs.

Trend

Composite score over time

Selected run

Module score breakdown

System spec

Selected run metadata

Select a run to inspect its notes, repo info, and benchmark configuration.

Compare visible runs

Composite scoreboard

Leaderboard

Published runs with full system context

# Label CPU OS Arch Cores RAM Composite Verdict Warnings Date
Loading benchmark history…

Module detail

Per-module results for the selected run

Module Score Ops / sec Wall time Flags Notes
Select a run to inspect module detail.