mightsleep commented on issue #11776: URL: https://github.com/apache/arrow-rs/issues/11776#issuecomment-6324648800
Yes hello, measured the x86 side on a Zen 5 laptop (Ryzen AI 5 340), current loop vs `count_ones_bytes`. your expectation holds, and the default x86-64 target is slower too: | target-cpu | 8 KB | 16 KB | 8 MB | |---|---|---|---| | x86-64 | 426 ns → 770 ns (+81%) | 851 ns → 1538 ns (+81%) | 436 µs → 788 µs (+81%) | | x86-64-v3 | 122 ns → 219 ns (+79%) | 245 ns → 437 ns (+79%) | 126 µs → 224 µs (+78%) | | native (znver5) | 42 ns → 75 ns (+77%) | 85 ns → 150 ns (+76%) | 63 µs → 85 µs (+33%) | So gating it on `target_arch = "aarch64"` looks right to me. -- 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]
