scovich commented on PR #7276: URL: https://github.com/apache/arrow-rs/pull/7276#issuecomment-2721859501
Finally got some reasonably stable benchmark results using an EC2 m6i.8xlarge instance, rustc 1.85.0. It uncovered one issue with the `append` helpers I had introduced. After addressing that, we now have: |benchmark|run1|run2|run3|run4|run5| |-|-|-|-|-|-| |small_bench_primitive|--noise--|--noise--|--noise--|--noise--|--noise--| |large_bench_primitive|2.27% faster|2.27% faster|1.65% faster|2.71% faster|1.36% faster| |small_bench_list|--noise--|--noise--|3.69% faster|2.71% faster|6.02% faster| I don't know why my changes should have caused a speedup, but at least there's no slowdown. Generated by: <details> ```sh # 5 runs against upstream main branch git checkout 82c2d5f4c for i in $(seq 5); do cargo bench --bench json_reader -- --save-baseline main$i; done # 5 runs against this PR git switch json-ignore-type-conflicts-option for i in $(seq 5); do cargo bench --bench json_reader -- --save-baseline feature$i; done # compare the run results for i in $(seq 5); do cargo bench --bench json_reader -- --load-baseline feature$i --baseline main$i; done ``` (see https://bheisler.github.io/criterion.rs/book/user_guide/command_line_options.html#baselines) </details> -- 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: github-unsubscr...@arrow.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
