Performance

Executed against commit 0bf47bec129ad3f2ae42ffcc73c031b416df6dbe

Spec: AMD Ryzen 7 5800X 8-Core, 32GB, GeForce RTX 3070

Rows manipulations benchmark:

npm run build:prod -w stress-app

cargo run --release -p stress -- --run table-ops --out benchmark_results/results.json

Median per op — 1k table (p50, ms)

Op	Rows	Ops Emitted	Total	Pre-apply	JS	Flush	Translate	Command	Layout	Bevy
create	0	4001	35.532	6.855	5.880	2.359	1.573	17.221	10.294	27.533
append1	1000	5	2.717	0.884	0.472	0.026	0.013	0.549	1.215	1.767
append1k	1001	4001	37.781	7.236	5.649	2.080	1.755	17.395	11.199	28.533
insert1	1000	5	3.859	2.363	0.675	0.029	0.015	0.547	1.178	1.726
insertEvery2nd	1001	2001	20.494	4.157	2.647	1.031	1.005	8.844	6.379	15.277
updateText1	1000	1	2.410	0.749	0.508	0.017	0.002	0.368	1.215	1.608
updateTextEvery2nd	1000	500	14.139	2.565	1.390	0.214	0.046	7.011	4.732	11.673
updateColor1	1000	1	1.873	0.937	0.575	0.017	0.010	0.341	0.562	0.903
updateColorEvery2nd	1000	500	4.173	2.860	1.853	0.369	0.452	0.385	0.519	0.880
updateTextColor1	1000	1	1.991	0.975	0.523	0.013	0.008	0.323	0.526	0.885
updateTextColorEvery2nd	1000	500	4.155	2.704	2.027	0.398	0.484	0.345	0.528	0.874
swap1	1000	997	6.316	4.412	2.754	0.226	0.325	0.391	1.135	1.519
swapEvery2nd	1000	500	4.058	2.564	0.762	0.131	0.201	0.379	1.252	1.616
remove1	1000	2	2.488	0.812	0.585	0.011	0.008	0.455	1.177	1.643
removeEvery2nd	999	500	4.500	0.844	0.716	0.106	0.566	1.973	0.956	2.876
clear	1000	1001	5.584	1.010	0.866	0.184	1.039	2.859	0.595	3.376

Op	Rows	Ops Emitted	Total	Pre-apply	JS	Flush	Translate	Command	Layout	Bevy
create	0	40001	512.458	212.153	211.947	19.124	12.865	171.523	116.635	287.014
append1	10000	5	25.809	6.760	3.696	0.031	0.019	2.856	16.631	19.522
append1k	10001	4001	68.300	20.899	8.738	2.032	1.831	19.224	25.906	45.300
insert1	10000	5	28.476	5.866	4.795	0.031	0.024	2.873	19.919	22.749
insertEvery2nd	10001	20001	209.431	37.634	28.336	10.214	8.588	87.454	75.246	161.702
updateText1	10000	1	24.239	6.132	5.110	0.018	0.002	1.378	15.904	17.377
updateTextEvery2nd	10000	5000	144.729	23.324	17.594	1.944	0.401	68.152	53.439	121.359
updateColor1	10000	1	30.160	21.226	6.367	0.020	0.013	1.384	8.710	10.098
updateColorEvery2nd	10000	5000	50.424	36.744	23.959	3.860	3.928	1.798	7.849	9.552
updateTextColor1	10000	1	31.426	21.746	6.450	0.019	0.013	1.195	8.344	9.505
updateTextColorEvery2nd	10000	5000	38.683	25.464	22.787	2.844	3.929	1.872	8.321	10.192
swap1	10000	9997	296.052	275.031	267.075	2.405	2.773	2.355	16.582	18.772
swapEvery2nd	10000	5000	43.390	22.516	10.010	1.092	1.780	1.998	17.220	19.220
remove1	10000	2	26.969	6.414	5.734	0.013	0.012	2.736	17.533	20.334
removeEvery2nd	9999	5000	54.743	20.908	7.455	0.992	4.260	19.777	10.091	29.888
clear	10000	10001	48.224	6.450	8.964	1.809	8.885	27.417	5.704	32.869

All timings are the median (p50) over the samples, in milliseconds.

Column	Meaning
Op	The operation under test (create, swap1, removeEvery2nd, …).
Rows	Table row count when the op ran (its precondition).
Ops Emitted	Size of the flushed op batch React produced for one occurrence of this op.
Total	End-to-end wall time, event trigger → post-layout on the frame the batch applied. Equals `Pre-apply + Translate + Bevy`.
Pre-apply	Trigger → Bevy starts applying the batch. Covers the JS round-trip + inter-thread scheduling. Contains JS.
JS	React reconcile + build the op batch + the `op_flush` call (measured on the JS thread). Subset of Pre-apply; contains Flush.
Flush	The `op_flush` native call alone = `JSON.stringify` of the batch + its `serde_json` decode. Subset of JS.
Translate	`apply_js_ops` walks the op batch → queues ECS commands (Bevy side).
Command	Execute the queued ECS commands + UI prepare/content, before layout.
Layout	`bevy_ui` layout: taffy solve + transform/clip propagation.
Bevy	Apply done → post-layout, same frame. Full post-translate Bevy wall time; ≈ `Command + Layout`.