pitrou commented on PR #39403:
URL: https://github.com/apache/arrow/pull/39403#issuecomment-1932082904
This PR *decreases* performance here (AMD Ryzen 9 3900X, gcc 12.3.0):
* before:
```
--------------------------------------------------------------------------------------------------------------------------------------------
Benchmark
Time CPU Iterations UserCounters...
--------------------------------------------------------------------------------------------------------------------------------------------
ReadLevels_Rle/MaxLevel:1/NumLevels:8096/BatchSize:1024/LevelRepeatCount:1
2325 ns 2326 ns 298740 bytes_per_second=6.48372Gi/s
items_per_second=3.48092G/s
ReadLevels_Rle/MaxLevel:1/NumLevels:8096/BatchSize:1024/LevelRepeatCount:7
8060 ns 8059 ns 87115 bytes_per_second=1.87109Gi/s
items_per_second=1.00453G/s
ReadLevels_Rle/MaxLevel:1/NumLevels:8096/BatchSize:1024/LevelRepeatCount:1024
668 ns 670 ns 1046591 bytes_per_second=22.5109Gi/s
items_per_second=12.0854G/s
ReadLevels_Rle/MaxLevel:1/NumLevels:8096/BatchSize:2048/LevelRepeatCount:1
2122 ns 2123 ns 330590 bytes_per_second=7.10169Gi/s
items_per_second=3.81269G/s
ReadLevels_Rle/MaxLevel:3/NumLevels:8096/BatchSize:1024/LevelRepeatCount:1
1948 ns 1949 ns 356752 bytes_per_second=7.7368Gi/s
items_per_second=4.15367G/s
ReadLevels_Rle/MaxLevel:3/NumLevels:8096/BatchSize:2048/LevelRepeatCount:1
1777 ns 1778 ns 395234 bytes_per_second=8.48193Gi/s
items_per_second=4.5537G/s
ReadLevels_Rle/MaxLevel:3/NumLevels:8096/BatchSize:1024/LevelRepeatCount:7
7568 ns 7568 ns 91365 bytes_per_second=1.99251Gi/s
items_per_second=1.06972G/s
ReadLevels_BitPack/MaxLevel:1/NumLevels:8096/BatchSize:1024/LevelRepeatCount:1
1252 ns 1257 ns 560376 bytes_per_second=11.9933Gi/s
items_per_second=6.43886G/s
ReadLevels_BitPack/MaxLevel:1/NumLevels:8096/BatchSize:1024/LevelRepeatCount:7
1257 ns 1263 ns 556189 bytes_per_second=11.9407Gi/s
items_per_second=6.41061G/s
ReadLevels_BitPack/MaxLevel:1/NumLevels:8096/BatchSize:1024/LevelRepeatCount:1024
1316 ns 1322 ns 550602 bytes_per_second=11.4078Gi/s
items_per_second=6.12452G/s
ReadLevels_BitPack/MaxLevel:1/NumLevels:8096/BatchSize:2048/LevelRepeatCount:1
1300 ns 1306 ns 517873 bytes_per_second=11.5467Gi/s
items_per_second=6.19908G/s
ReadLevels_BitPack/MaxLevel:3/NumLevels:8096/BatchSize:1024/LevelRepeatCount:1
1312 ns 1318 ns 456593 bytes_per_second=11.4407Gi/s
items_per_second=6.14218G/s
ReadLevels_BitPack/MaxLevel:3/NumLevels:8096/BatchSize:2048/LevelRepeatCount:1
1250 ns 1256 ns 562519 bytes_per_second=12.0082Gi/s
items_per_second=6.44683G/s
ReadLevels_BitPack/MaxLevel:3/NumLevels:8096/BatchSize:1024/LevelRepeatCount:7
1301 ns 1307 ns 519667 bytes_per_second=11.5369Gi/s
items_per_second=6.19383G/s
```
* after:
```
--------------------------------------------------------------------------------------------------------------------------------------------
Benchmark
Time CPU Iterations UserCounters...
--------------------------------------------------------------------------------------------------------------------------------------------
ReadLevels_Rle/MaxLevel:1/NumLevels:8096/BatchSize:1024/LevelRepeatCount:1
2563 ns 2566 ns 273640 bytes_per_second=5.87793Gi/s
items_per_second=3.15569G/s
ReadLevels_Rle/MaxLevel:1/NumLevels:8096/BatchSize:1024/LevelRepeatCount:7
8134 ns 8135 ns 85865 bytes_per_second=1.85376Gi/s
items_per_second=995.23M/s
ReadLevels_Rle/MaxLevel:1/NumLevels:8096/BatchSize:1024/LevelRepeatCount:1024
725 ns 726 ns 956867 bytes_per_second=20.7575Gi/s
items_per_second=11.1441G/s
ReadLevels_Rle/MaxLevel:1/NumLevels:8096/BatchSize:2048/LevelRepeatCount:1
2304 ns 2306 ns 302796 bytes_per_second=6.53916Gi/s
items_per_second=3.51068G/s
ReadLevels_Rle/MaxLevel:3/NumLevels:8096/BatchSize:1024/LevelRepeatCount:1
2087 ns 2089 ns 334483 bytes_per_second=7.21913Gi/s
items_per_second=3.87574G/s
ReadLevels_Rle/MaxLevel:3/NumLevels:8096/BatchSize:2048/LevelRepeatCount:1
1849 ns 1851 ns 377499 bytes_per_second=8.14629Gi/s
items_per_second=4.37351G/s
ReadLevels_Rle/MaxLevel:3/NumLevels:8096/BatchSize:1024/LevelRepeatCount:7
8214 ns 8216 ns 85257 bytes_per_second=1.83534Gi/s
items_per_second=985.339M/s
ReadLevels_BitPack/MaxLevel:1/NumLevels:8096/BatchSize:1024/LevelRepeatCount:1
1370 ns 1372 ns 508348 bytes_per_second=10.9904Gi/s
items_per_second=5.90044G/s
ReadLevels_BitPack/MaxLevel:1/NumLevels:8096/BatchSize:1024/LevelRepeatCount:7
1366 ns 1368 ns 507420 bytes_per_second=11.024Gi/s
items_per_second=5.91848G/s
ReadLevels_BitPack/MaxLevel:1/NumLevels:8096/BatchSize:1024/LevelRepeatCount:1024
1367 ns 1369 ns 510455 bytes_per_second=11.0135Gi/s
items_per_second=5.91285G/s
ReadLevels_BitPack/MaxLevel:1/NumLevels:8096/BatchSize:2048/LevelRepeatCount:1
1377 ns 1379 ns 507504 bytes_per_second=10.9347Gi/s
items_per_second=5.87051G/s
ReadLevels_BitPack/MaxLevel:3/NumLevels:8096/BatchSize:1024/LevelRepeatCount:1
1367 ns 1369 ns 511657 bytes_per_second=11.0171Gi/s
items_per_second=5.91476G/s
ReadLevels_BitPack/MaxLevel:3/NumLevels:8096/BatchSize:2048/LevelRepeatCount:1
1332 ns 1333 ns 524574 bytes_per_second=11.3119Gi/s
items_per_second=6.07305G/s
ReadLevels_BitPack/MaxLevel:3/NumLevels:8096/BatchSize:1024/LevelRepeatCount:7
1373 ns 1374 ns 510605 bytes_per_second=10.9713Gi/s
items_per_second=5.89016G/s
```
--
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]