mrk-andreev commented on code in PR #48501:
URL: https://github.com/apache/spark/pull/48501#discussion_r1844407932
##########
sql/core/benchmarks/CollationBenchmark-jdk21-results.txt:
##########
@@ -1,54 +1,88 @@
-OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.5.0-1025-azure
-AMD EPYC 7763 64-Core Processor
+OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.8.0-1017-aws
+Intel(R) Xeon(R) CPU E5-2686 v4 @ 2.30GHz
collation unit benchmarks - equalsFunction: Best Time(ms) Avg Time(ms)
Stdev(ms) Rate(M/s) Per Row(ns) Relative time
--------------------------------------------------------------------------------------------------------------------------
-UTF8_BINARY 1349 1349
0 0.1 13485.4 1.0X
-UTF8_LCASE 3559 3561
3 0.0 35594.3 2.6X
-UNICODE 17580 17589
12 0.0 175803.6 13.0X
-UNICODE_CI 17210 17212
2 0.0 172100.2 12.8X
+UTF8_BINARY 2220 2223
5 0.0 22197.0 1.0X
+UTF8_LCASE 4949 4950
2 0.0 49488.1 2.2X
+UNICODE 28172 28198
36 0.0 281721.0 12.7X
+UNICODE_CI 28233 28308
106 0.0 282328.2 12.7X
-OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.5.0-1025-azure
-AMD EPYC 7763 64-Core Processor
+OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.8.0-1017-aws
+Intel(R) Xeon(R) CPU E5-2686 v4 @ 2.30GHz
collation unit benchmarks - compareFunction: Best Time(ms) Avg Time(ms)
Stdev(ms) Rate(M/s) Per Row(ns) Relative time
---------------------------------------------------------------------------------------------------------------------------
-UTF8_BINARY 1740 1741
1 0.1 17398.8 1.0X
-UTF8_LCASE 2630 2632
3 0.0 26301.0 1.5X
-UNICODE 16732 16743
16 0.0 167319.7 9.6X
-UNICODE_CI 16482 16492
14 0.0 164819.7 9.5X
+UTF8_BINARY 2731 2733
2 0.0 27313.6 1.0X
+UTF8_LCASE 4611 4619
11 0.0 46111.4 1.7X
+UNICODE 28149 28211
88 0.0 281486.8 10.3X
+UNICODE_CI 27535 27597
89 0.0 275348.4 10.1X
-OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.5.0-1025-azure
-AMD EPYC 7763 64-Core Processor
+OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.8.0-1017-aws
+Intel(R) Xeon(R) CPU E5-2686 v4 @ 2.30GHz
collation unit benchmarks - hashFunction: Best Time(ms) Avg Time(ms)
Stdev(ms) Rate(M/s) Per Row(ns) Relative time
------------------------------------------------------------------------------------------------------------------------
-UTF8_BINARY 2808 2808
0 0.0 28082.3 1.0X
-UTF8_LCASE 5412 5413
1 0.0 54123.5 1.9X
-UNICODE 70755 70787
44 0.0 707553.4 25.2X
-UNICODE_CI 57639 57669
43 0.0 576390.0 20.5X
+UTF8_BINARY 4603 4618
22 0.0 46031.3 1.0X
+UTF8_LCASE 9510 9518
11 0.0 95097.7 2.1X
+UNICODE 135718 135786
97 0.0 1357176.2 29.5X
+UNICODE_CI 113715 113819
148 0.0 1137145.8 24.7X
-OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.5.0-1025-azure
-AMD EPYC 7763 64-Core Processor
+OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.8.0-1017-aws
+Intel(R) Xeon(R) CPU E5-2686 v4 @ 2.30GHz
collation unit benchmarks - contains: Best Time(ms) Avg Time(ms)
Stdev(ms) Rate(M/s) Per Row(ns) Relative time
------------------------------------------------------------------------------------------------------------------------
-UTF8_BINARY 9356 9357
0 0.0 93564.9 1.0X
-UTF8_LCASE 24106 24129
33 0.0 241055.3 2.6X
-UNICODE 368428 369053
883 0.0 3684284.1 39.4X
-UNICODE_CI 417361 418242
1246 0.0 4173613.9 44.6X
+UTF8_BINARY 14957 14959
4 0.0 149568.0 1.0X
+UTF8_LCASE 49434 49617
259 0.0 494340.1 3.3X
+UNICODE 920911 921348
618 0.0 9209111.3 61.6X
+UNICODE_CI 988411 988822
582 0.0 9884107.7 66.1X
-OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.5.0-1025-azure
-AMD EPYC 7763 64-Core Processor
+OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.8.0-1017-aws
+Intel(R) Xeon(R) CPU E5-2686 v4 @ 2.30GHz
collation unit benchmarks - startsWith: Best Time(ms) Avg Time(ms)
Stdev(ms) Rate(M/s) Per Row(ns) Relative time
------------------------------------------------------------------------------------------------------------------------
-UTF8_BINARY 10941 10943
2 0.0 109411.5 1.0X
-UTF8_LCASE 20041 20058
24 0.0 200410.1 1.8X
-UNICODE 364296 365610
1859 0.0 3642958.8 33.3X
-UNICODE_CI 424306 424888
823 0.0 4243062.7 38.8X
+UTF8_BINARY 17517 17523
9 0.0 175173.1 1.0X
+UTF8_LCASE 33605 33643
54 0.0 336053.8 1.9X
+UNICODE 916945 917042
139 0.0 9169445.3 52.3X
+UNICODE_CI 992431 992565
189 0.0 9924305.9 56.7X
-OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.5.0-1025-azure
-AMD EPYC 7763 64-Core Processor
+OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.8.0-1017-aws
+Intel(R) Xeon(R) CPU E5-2686 v4 @ 2.30GHz
collation unit benchmarks - endsWith: Best Time(ms) Avg Time(ms)
Stdev(ms) Rate(M/s) Per Row(ns) Relative time
------------------------------------------------------------------------------------------------------------------------
-UTF8_BINARY 10551 10556
7 0.0 105511.7 1.0X
-UTF8_LCASE 20294 20300
9 0.0 202943.7 1.9X
-UNICODE 384070 384554
684 0.0 3840704.6 36.4X
-UNICODE_CI 441935 442184
352 0.0 4419351.4 41.9X
+UTF8_BINARY 17482 17485
5 0.0 174819.0 1.0X
+UTF8_LCASE 33890 33925
49 0.0 338901.6 1.9X
+UNICODE 929570 930339
1087 0.0 9295702.4 53.2X
+UNICODE_CI 1002779 1003427
916 0.0 10027792.9 57.4X
+
+OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.8.0-1017-aws
+Intel(R) Xeon(R) CPU E5-2686 v4 @ 2.30GHz
+collation unit benchmarks - initCap using impl execICU: Best Time(ms) Avg
Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative time
+--------------------------------------------------------------------------------------------------------------------------------------
+UNICODE 10
11 1 9.7 102.6 1.0X
+UNICODE_CI 12
12 1 8.4 118.8 1.2X
+
+OpenJDK 64-Bit Server VM 21.0.5+11-LTS on Linux 6.8.0-1017-aws
+Intel(R) Xeon(R) CPU E5-2686 v4 @ 2.30GHz
+collation unit benchmarks - initCap using impl execBinaryICU: Best Time(ms)
Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative time
+--------------------------------------------------------------------------------------------------------------------------------------------
+UTF8_BINARY 2
2 0 64.6 15.5 1.0X
Review Comment:
As we can see, this benchmark is sensitive to CPU clock speed. During my
latest measurements on an `Intel(R) Xeon(R) Platinum 8252C CPU @ 3.80GHz` (AWS
m5zn.xlarge), the stdev for some measurements - along with others - dropped to
zero or one.
I suggest adding more decimal places to the results in a separate PR.
--
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: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
