HaoYang670 commented on PR #2127:
URL: https://github.com/apache/arrow-rs/pull/2127#issuecomment-1194749573
> In the past I've noticed significant regressions when I've accidentally
reordered operations around bound checks in such a way that LLVM cannot elide
them. Not sure if that applies in this case though... Capacity reservation
would be another likely suspect
Amazing @tustvold, you hit it! After reordering the 2 lines in
`BooleanBuilder::append_slice`, the performance regression disappeared!
```
bench_bool/bench_bool time: [259.80 us 260.93 us 262.23 us]
thrpt: [1.8621 GiB/s 1.8713 GiB/s 1.8795 GiB/s]
change:
time: [-10.518% -10.061% -9.5849%] (p = 0.00 <
0.05)
thrpt: [+10.601% +11.186% +11.754%]
Performance has improved.
Found 7 outliers among 100 measurements (7.00%)
2 (2.00%) high mild
5 (5.00%) high severe
```
--
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]
