viirya commented on PR #3361:
URL: https://github.com/apache/arrow-rs/pull/3361#issuecomment-1357285491
Updated: By implementing some functions of `Write`, most regressions
disappear.
Still regression ~25%: string bloom filter, string dictionary with bloom
filter, string non-null with bloom filter.
```
write_batch primitive/4096 values primitive
time: [625.94 µs 626.93 µs 628.10 µs]
thrpt: [281.04 MiB/s 281.57 MiB/s 282.01 MiB/s]
change:
time: [-0.0703% +0.2373% +0.5201%] (p = 0.12 >
0.05)
thrpt: [-0.5174% -0.2367% +0.0704%]
No change in performance detected.
Found 7 outliers among 100 measurements (7.00%)
6 (6.00%) high mild
1 (1.00%) high severe
write_batch primitive/4096 values primitive with bloom filter
time: [3.9652 ms 3.9857 ms 4.0075 ms]
thrpt: [44.048 MiB/s 44.289 MiB/s 44.518 MiB/s]
change:
time: [-3.7486% -3.1900% -2.6823%] (p = 0.00 <
0.05)
thrpt: [+2.7562% +3.2951% +3.8946%]
Performance has improved.
Found 15 outliers among 100 measurements (15.00%)
15 (15.00%) high mild
write_batch primitive/4096 values primitive non-null
time: [537.44 µs 540.51 µs 543.10 µs]
thrpt: [318.73 MiB/s 320.26 MiB/s 322.09 MiB/s]
change:
time: [-2.6202% -1.7441% -0.7530%] (p = 0.00 <
0.05)
thrpt: [+0.7587% +1.7751% +2.6907%]
Change within noise threshold.
write_batch primitive/4096 values primitive non-null with bloom filter
time: [3.8815 ms 3.8926 ms 3.9050 ms]
thrpt: [44.328 MiB/s 44.470 MiB/s 44.597 MiB/s]
change:
time: [-2.3063% -1.9639% -1.6055%] (p = 0.00 <
0.05)
thrpt: [+1.6317% +2.0032% +2.3608%]
Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
3 (3.00%) high mild
6 (6.00%) high severe
write_batch primitive/4096 values bool
time: [51.376 µs 52.002 µs 52.541 µs]
thrpt: [21.926 MiB/s 22.154 MiB/s 22.424 MiB/s]
change:
time: [-40.319% -39.705% -39.077%] (p = 0.00 <
0.05)
thrpt: [+64.143% +65.851% +67.558%]
Performance has improved.
Found 23 outliers among 100 measurements (23.00%)
23 (23.00%) high mild
write_batch primitive/4096 values bool non-null
time: [30.089 µs 30.094 µs 30.099 µs]
thrpt: [22.053 MiB/s 22.056 MiB/s 22.060 MiB/s]
change:
time: [-59.123% -59.067% -58.987%] (p = 0.00 <
0.05)
thrpt: [+143.83% +144.30% +144.64%]
Performance has improved.
Found 10 outliers among 100 measurements (10.00%)
4 (4.00%) high mild
6 (6.00%) high severe
write_batch primitive/4096 values string
time: [275.84 µs 278.10 µs 280.72 µs]
thrpt: [283.68 MiB/s 286.35 MiB/s 288.70 MiB/s]
change:
time: [-0.0601% +1.0398% +2.2510%] (p = 0.08 >
0.05)
thrpt: [-2.2014% -1.0291% +0.0602%]
No change in performance detected.
Found 12 outliers among 100 measurements (12.00%)
12 (12.00%) high severe
write_batch primitive/4096 values string with bloom filter
time: [966.97 µs 988.71 µs 1.0155 ms]
thrpt: [78.417 MiB/s 80.545 MiB/s 82.356 MiB/s]
change:
time: [+20.632% +24.001% +27.467%] (p = 0.00 <
0.05)
thrpt: [-21.548% -19.356% -17.103%]
Performance has regressed.
Found 16 outliers among 100 measurements (16.00%)
6 (6.00%) low mild
2 (2.00%) high mild
8 (8.00%) high severe
write_batch primitive/4096 values string dictionary
time: [150.19 µs 151.71 µs 153.53 µs]
thrpt: [313.56 MiB/s 317.32 MiB/s 320.54 MiB/s]
change:
time: [+2.4272% +3.6178% +4.7690%] (p = 0.00 <
0.05)
thrpt: [-4.5519% -3.4915% -2.3697%]
Performance has regressed.
write_batch primitive/4096 values string dictionary with bloom filter
time: [549.69 µs 559.42 µs 567.68 µs]
thrpt: [84.805 MiB/s 86.056 MiB/s 87.579 MiB/s]
change:
time: [+23.015% +26.986% +31.032%] (p = 0.00 <
0.05)
thrpt: [-23.683% -21.251% -18.709%]
Performance has regressed.
write_batch primitive/4096 values string non-null
time: [339.09 µs 342.80 µs 346.21 µs]
thrpt: [227.20 MiB/s 229.46 MiB/s 231.97 MiB/s]
change:
time: [+1.3603% +2.0031% +2.7986%] (p = 0.00 <
0.05)
thrpt: [-2.7224% -1.9638% -1.3421%]
Performance has regressed.
Found 23 outliers among 100 measurements (23.00%)
23 (23.00%) high mild
Benchmarking write_batch primitive/4096 values string non-null with bloom
filter: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase
target time to 5.4s, enable flat sampling, or reduce sample count to 60.
write_batch primitive/4096 values string non-null with bloom filter
time: [1.1739 ms 1.1742 ms 1.1746 ms]
thrpt: [66.968 MiB/s 66.988 MiB/s 67.004 MiB/s]
change:
time: [+21.914% +25.172% +28.517%] (p = 0.00 <
0.05)
thrpt: [-22.189% -20.110% -17.975%]
Performance has regressed.
Found 9 outliers among 100 measurements (9.00%)
4 (4.00%) high mild
5 (5.00%) high severe
write_batch nested/4096 values primitive list
time: [935.44 µs 942.05 µs 948.25 µs]
thrpt: [172.82 MiB/s 173.96 MiB/s 175.19 MiB/s]
change:
time: [-0.6966% -0.0169% +0.6661%] (p = 0.96 >
0.05)
thrpt: [-0.6617% +0.0169% +0.7015%]
No change in performance detected.
Benchmarking write_batch nested/4096 values primitive list non-null: Warming
up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase
target time to 5.8s, enable flat sampling, or reduce sample count to 60.
write_batch nested/4096 values primitive list non-null
time: [1.1124 ms 1.1230 ms 1.1339 ms]
thrpt: [168.05 MiB/s 169.68 MiB/s 171.30 MiB/s]
change:
time: [+0.2012% +1.2888% +2.3253%] (p = 0.02 <
0.05)
thrpt: [-2.2725% -1.2724% -0.2008%]
Change within noise threshold.
```
--
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]
