mbutrovich commented on PR #11425: URL: https://github.com/apache/arrow-rs/pull/11425#issuecomment-6064193328
Thanks for the thorough review @andygrove. > **`test_i256_fuzz` never generates negative operands.** I filed #11452 for this. > **`divide_and_round` in arrow-arith could be the first caller in this repo.** I added `multiply_fixed_point` benchmarks to #11427, and I filed #11453 for the change itself, which depends on this PR. > **Generic decimal code can't reach the fused method.** #11426 already benchmarks `make_downscaler`, and I added `div` cases to #11427 whose scaled dividend overflows, so they reach `scaled_div`. I filed #11454 for the trait methods. > **A benchmark would show the gain.** The `multiply_fixed_point` benchmarks in #11427 cover this. They will show the gain once `divide_and_round` uses `wrapping_div_rem`. > **`div_rem` itself has room to speed up.** I filed #11451 for the `i128` fast path and the missing `#[inline]` attributes. The `Decimal256` cases in #11426 and #11427 can measure both. -- 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]
