About
One person, three rigs, every run published.
Why this exists
Most tool comparisons are written by someone selling one of the tools. The rest quote a number with no machine, no versions, and no workload behind it. I wanted the page barefeats used to publish for Mac hardware, but for the choices I actually have to make at work: the linter, the bundler, the cache, the desktop shell, the map renderer, the tile server.
Who runs it
I'm John Carmack. I write Rust and TypeScript for a living. I run every benchmark here myself, on hardware I own or rent, with a runner you can read. No sponsor. No affiliation with any tool on this site.
The rigs
Three so far. Most runs come off an M2 Max MacBook Pro with 96 GB. A BOSGAME mini PC (Ryzen 7 6800H, Radeon 680M, Windows 11) ran MapLibre vs Mapbox and the Windows half of Tauri vs Electron. Redis vs Valkey vs Dragonfly ran on AWS Graviton3: a c7g.metal engine with a c7g.4xlarge client. Every result page names its rig, and numbers from different rigs never share a chart.
Independence
Hosting comes out of my pocket, and later out of ads. If a sponsor ever pays for a bigger rig, that sponsor gets named on every result that ran on it. Nobody but me picks what gets benchmarked or how.
Logos
Product logos appear unmodified to identify the products under test and imply no endorsement. Varnish is a registered trademark of Varnish Software AB. NGINX is a trademark of F5, Inc. The Vinyl Cache logo is CC BY 4.0 Rhubarbe.design. ESLint and Biome logos belong to their respective projects. The Vite and esbuild logos are MIT-licensed from their respective project repositories. Redis is a registered trademark of Redis Ltd. Any rights therein are reserved to Redis Ltd. Valkey is a trademark of LF Projects, LLC. Dragonfly is a trademark of DragonflyDB Ltd.
Corrections
Think a result is wrong? Open an issue in the repository with your rig and your samples. Corrections go up as new runs. The original stays, with a note.