🏎️ does JIT go brrr?

A performance dashboard tracking CPython's JIT vs. Interpreter benchmarks.

View live at doesjitgobrrr.com or isthejitfasteryet.com.

Architecture

The dashboard is served by FastAPI Cloud, but the app is operationally static:

• FastAPI serves the built Solid/Vite frontend from api/static.
• Benchmark data is generated into JSON files under frontend/public/data.
• The frontend reads /data/*.json static assets directly. There is no runtime database query path.

New benchmark data enters the app through the existing nightly benchmark flow: pyperf_bench finishes its nightly run, triggers this repository's Update Dashboard Data workflow with a benchmark_completed repository_dispatch, and this repo appends newly discovered benchmark runs to the cached static data blob. The workflow then rewrites the public /data JSON files, rebuilds the frontend, copies the build to api/static, and deploys the refreshed static bundle to FastAPI Cloud. After that deployment, a dashboard refresh revalidates the /data files.

Development

Prerequisites

• Node.js 20+ (for frontend development)
• Python 3.13+ with uv (for data generation and FastAPI Cloud deploys)

Local Frontend Development

Generate static data, then run the frontend:

cd api
uv run python generate_static_data.py --out ../frontend/public/data

cd ../frontend
npm ci
npm run dev

For a faster local smoke run with real GitHub data, process only the newest missing benchmark pairs:

cd api
uv run python generate_static_data.py --max-pairs 10 --out ../frontend/public/data

If you already have api/.static-data-cache.json and only need to rewrite the public static assets from it:

cd api
uv run python generate_static_data.py --skip-fetch --out ../frontend/public/data

Deployment

Deployment is handled by .github/workflows/fastapicloud-deploy.yml:

1. Generate static dashboard data into frontend/public/data.
2. Build the Solid/Vite frontend.
3. Copy frontend/dist to api/static.
4. Deploy the FastAPI app to FastAPI Cloud.

Data Refresh

Benchmark data is refreshed via GitHub Actions in .github/workflows/update-data.yml. The workflow keeps the existing external trigger: the nightly benchmark workflow in pyperf_bench sends a benchmark_completed repository dispatch here when new results are ready.

The generator restores api/.static-data-cache.json from the GitHub Actions cache, treats it as the appendable source-of-truth blob, skips benchmark directories already present in that blob, appends newly discovered runs, and then writes public static JSON:

• manifest.json for available dates and metadata
• machines.json and events.json
• summary-7.json, summary-30.json, and summary-all.json for chart data
• runs/YYYY-MM-DD.json for detail pages

Change annotations remain authored in api/perf_events.yaml. The static data generator validates and sorts that YAML on every data refresh or deploy, then writes it to /data/events.json for the chart's Changes toggle. The existing Suggest annotations workflow still opens PRs against api/perf_events.yaml; merging one of those PRs to main triggers the normal FastAPI Cloud deploy, which rebuilds events.json.

Contributing

Want to contribute benchmark data from your own hardware? See CONTRIBUTING.md for instructions on adding your machines to the dashboard.

License

MIT