avantgardnerio commented on PR #7192:
URL:
https://github.com/apache/arrow-datafusion/pull/7192#issuecomment-1689158528
> Reported performance results
I'd like to reiterate that this PR is really about using constant memory
(which it does), not increasing throughput, but here's some throughput numbers
I just got by running `sudo cargo flamegraph --bench topk_aggregate -- --bench`:
```
Benchmarking aggregate 10000000 time-series rows: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase
target time to 11.3s, or reduce sample count to 40.
aggregate 10000000 time-series rows
time: [132.74 ms 134.27 ms 135.80 ms]
change: [-0.2640% +1.2977% +2.8872%] (p = 0.10 >
0.05)
No change in performance detected.
Benchmarking aggregate 10000000 worst-case rows: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase
target time to 14.4s, or reduce sample count to 30.
aggregate 10000000 worst-case rows
time: [148.49 ms 150.24 ms 152.11 ms]
change: [+9.1483% +10.681% +12.374%] (p = 0.00 <
0.05)
Performance has regressed.
Found 6 outliers among 100 measurements (6.00%)
4 (4.00%) high mild
2 (2.00%) high severe
top k=10 aggregate 10000000 time-series rows
time: [42.050 ms 42.990 ms 43.940 ms]
change: [-86.190% -85.548% -84.859%] (p = 0.00 <
0.05)
Performance has improved.
Found 1 outliers among 100 measurements (1.00%)
1 (1.00%) high mild
Benchmarking top k=10 aggregate 10000000 worst-case rows: Warming up for
3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase
target time to 7.1s, or reduce sample count to 70.
top k=10 aggregate 10000000 worst-case rows
time: [70.896 ms 73.007 ms 75.216 ms]
change: [-93.860% -93.308% -92.702%] (p = 0.00 <
0.05)
Performance has improved.
Found 2 outliers among 100 measurements (2.00%)
2 (2.00%) high mild
```
--
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]
