Rich-T-kid commented on PR #11973: URL: https://github.com/apache/arrow-rs/pull/11973#issuecomment-6688574816
Final-head local A/B against `739f9cc34a` (separate release binaries, order-balanced H/B/B/H, 20 samples, 1 s warmup, 2 s measurement): | Benchmark | Base median | PR median | Delta | |---|---:|---:|---:| | mutable str | 2.560 ms | 2.516 ms | -1.7% | | mutable str nulls | 1.356 ms | 1.349 ms | -0.5% | | mutable str_view | 168.7 µs | 127.3 µs | -24.5% | | mutable str_view short slices | 10.47 µs | 10.86 µs | +3.7% | | mutable dict | 37.54 µs | 38.09 µs | +1.5% | | mutable dict short slices | 11.87 µs | 12.40 µs | +4.5% | | dict-overflow merge, 50% true | 50.84 µs | 50.77 µs | -0.1% | | dict-overflow merge, 50% null | 50.12 µs | 50.37 µs | +0.5% | The previously reported 20–48% dictionary-overflow merge regressions are not reproduced on the combined head. The remaining local slowdowns are confined to the short-slice microbenchmarks, so I am rerunning all affected hosted suites on `c493936373` below. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
