Measured locally. Not a GPU FPS claim. See docs/BENCHMARK_METHODOLOGY.md
Last measured 2026-08-13 on Intel(R) Core(TM) i5-6500T CPU @ 2.50GHz (4 logical CPUs, win32/x64, Node v24.14.1). Reproduce locally with pnpm benchmark — see docs/BENCHMARK_METHODOLOGY.md.
| Points | Fill | Init | LOD | Upload | Render | Pool churn | LOD level |
|---|---|---|---|---|---|---|---|
| 1,000 | 0.217ms | 0.647ms | 0.564ms | 0.148ms | 1.012ms | 2.781ms | raw |
| 10,000 | 1.584ms | 0.413ms | 1.905ms | 0.546ms | 0.326ms | 0.909ms | moderate |
| 100,000 | 9.695ms | 1.148ms | 3.939ms | 0.168ms | 0.013ms | 0.259ms | aggressive |
These are CPU-side data-path timings (typed-array fill, engine init, LOD selection, GPU buffer upload prep, render-graph compile, buffer-pool churn) from @rekhly/data and @rekhly/engine on this machine's CPU — not a GPU frame-rate claim, since actual FPS depends on the GPU, driver, and display the chart renders on.
Got any questions? I'm happy to help with products, licensing, and examples.