Closed adriencaccia closed 8 months ago
Comparing feat/single-build
(69c8309) with main
(fc75139)
⚡ 3
improvements
✅ 54
untouched benchmarks
Benchmark | main |
feat/single-build |
Change | |
---|---|---|---|---|
⚡ | Auto |
223.3 ns | 195.6 ns | +14.2% |
⚡ | Linear |
223.3 ns | 195.6 ns | +14.2% |
⚡ | Flat |
728.3 ns | 700.6 ns | +3.97% |
@art049 I made the changes, let me know if it's okay to merge. I don't really know what to make of the performance improvements of the criterion sampling modes benchmarks.
LGTM, @adriencaccia; those benches are relying on sleep. Thus we end up with syscall variance.
As pointed out by @Boshen in https://github.com/oxc-project/oxc/pull/2343, we can build all the benches in a single command.