mbutrovich commented on issue #11776: URL: https://github.com/apache/arrow-rs/issues/11776#issuecomment-6345432851
https://github.com/apache/arrow-rs/issues/11776#issuecomment-6337172176 > @mbutrovich Can you please run the benchmark on non-apple arm64 machine, sometimes apple perf can be very different especially with Neoverse v1 and v2 like AWS Graviton 3 and 4. On Neoverse V2 (Google Axion), `true_count` without nulls in #11818 is 1.87x faster at 65536 bits. It held with code layout controlled. The PR description has the other sizes and the amd64 results. @mightsleep, the 64-bit cases on amd64 are also faster (2.2x), and I can't explain it from the generated code. Would you be up for running `true_count(all_true, 64)` from the `boolean_array` bench on `main` and on #11818 under `perf stat` on your Zen 5 machine? -- 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]
