gstvg commented on PR #8967:
URL: https://github.com/apache/arrow-rs/pull/8967#issuecomment-3639707460
Just tested on AMD, also lower improvements:
<details>
`-C target-cpu=znver3`
```
take bool 512 time: [483.26 ns 483.65 ns 484.09 ns]
change: [−12.699% −12.572% −12.446%] (p = 0.00 <
0.05)
Performance has improved.
take bool 1024 time: [872.58 ns 873.03 ns 873.43 ns]
change: [−13.936% −13.860% −13.775%] (p = 0.00 <
0.05)
Performance has improved.
take bool 8192 time: [5.6960 µs 5.7022 µs 5.7096 µs]
change: [−23.644% −23.566% −23.486%] (p = 0.00 <
0.05)
Performance has improved.
```
`-C target-cpu=znver4`
```
take bool 512 time: [447.41 ns 447.57 ns 447.74 ns]
change: [−9.4770% −9.2941% −9.0973%] (p = 0.00 <
0.05)
Performance has improved.
take bool 1024 time: [787.11 ns 787.30 ns 787.49 ns]
change: [−12.403% −12.247% −12.099%] (p = 0.00 <
0.05)
Performance has improved.
take bool 8192 time: [5.6537 µs 5.6553 µs 5.6569 µs]
change: [−14.204% −14.055% −13.929%] (p = 0.00 <
0.05)
Performance has improved.
```
Lower gains in zen 4 than zen 3 is due to
https://github.com/llvm/llvm-project/issues/91370, targeting znver3 for Zen4
CPU improves a bit.
`-C target-cpu=znver5`
```
take bool 512 time: [325.74 ns 325.76 ns 325.79 ns]
change: [−6.9312% −6.6598% −6.4780%] (p = 0.00 <
0.05)
Performance has improved.
take bool 1024 time: [492.13 ns 492.42 ns 492.67 ns]
change: [−22.795% −22.746% −22.694%] (p = 0.00 <
0.05)
Performance has improved.
take bool 8192 time: [3.3653 µs 3.3656 µs 3.3658 µs]
change: [−25.208% −25.171% −25.137%] (p = 0.00 <
0.05)
Performance has improved.
```
</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: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
