Same world · four renderers

Game of Life runtime comparison

Four production builds render the same finite board, controls, presets, accessibility contract, and visual system. Lower timings are better; the fastest result in each row is highlighted.

Kineticadev · 98.3 KB gzip React19.2.7 · 62.3 KB gzip Compose HTML1.11.1 · 177.3 KB gzip Vanillabrowser DOM · 4.1 KB gzip

Measured performance

Median of 5 measured runs after 1 warmup; milliseconds and slowdown versus the fastest implementation per operation.

Operation	Kinetica	React	Compose HTML	Vanilla
Cold startup + 3,456-cell mount	66.90 ms1.88×	56.00 ms1.57×	107.40 ms3.02×	35.60 ms1.00×
Load 48-cell Pulsar preset	21.42 ms1.00×	21.87 ms1.02×	34.62 ms1.62×	27.18 ms1.27×
Advance Pulsar one generation	24.23 ms1.00×	28.43 ms1.17×	31.85 ms1.31×	24.82 ms1.02×
Populate 24% of the 3,456-cell board	140.39 ms1.18×	155.99 ms1.31×	140.28 ms1.18×	118.71 ms1.00×
Advance a randomized board one generation	121.97 ms1.04×	149.87 ms1.28×	128.78 ms1.10×	116.83 ms1.00×
Toggle one editable board cell	14.50 ms1.08×	18.93 ms1.41×	13.75 ms1.03×	13.41 ms1.00×
Clear the 48-cell Pulsar	21.00 ms1.04×	20.15 ms1.00×	33.28 ms1.65×	20.25 ms1.00×

Chromium 149.0.7827.55arm641280x9002026-07-16

Durations use the same trace parser as Kinetica’s framework benchmark: click dispatch through the last paint/commit. Interaction traces request reduced motion so the shared 160 ms cell-birth decoration cannot mask renderer work. Startup includes loading the production bundle and mounting 3,456 interactive cells. Randomized runs use the same cross-language seeded sequence at 24% density, producing a fresh but identical workload for every renderer. Results are machine-specific. Download raw samples and environment metadata.