[
https://issues.apache.org/jira/browse/FLINK-34148?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17808618#comment-17808618
]
Zakelly Lan commented on FLINK-34148:
-------------------------------------
After some investigation, I found it is this commit
[5d9d8748b6|https://github.com/apache/flink/commit/5d9d8748b64ff1a75964a5cd2857ab5061312b51]
that introduce the regression.
The result without this commit:
{code:java}
19:10:36 Benchmark (lengthStr) (type) Mode
Cnt Score Error Units
19:10:36 StringSerializationBenchmark.stringWrite 4 ascii thrpt
30 13269.950 ± 32.092 ops/ms
19:10:36 StringSerializationBenchmark.stringWrite 4 russian thrpt
30 11040.363 ± 47.226 ops/ms
19:10:36 StringSerializationBenchmark.stringWrite 4 chinese thrpt
30 9554.081 ± 38.286 ops/ms
19:10:36 StringSerializationBenchmark.stringWrite 128 ascii thrpt
30 1086.989 ± 3.005 ops/ms
19:10:36 StringSerializationBenchmark.stringWrite 128 russian thrpt
30 654.435 ± 1.718 ops/ms
19:10:36 StringSerializationBenchmark.stringWrite 128 chinese thrpt
30 502.264 ± 2.158 ops/ms
19:10:36 StringSerializationBenchmark.stringWrite 16384 ascii thrpt
30 6.041 ± 0.013 ops/ms
19:10:36 StringSerializationBenchmark.stringWrite 16384 russian thrpt
30 3.203 ± 0.008 ops/ms
19:10:36 StringSerializationBenchmark.stringWrite 16384 chinese thrpt
30 2.266 ± 0.007 ops/ms {code}
with this commit:
{code:java}
19:27:30 Benchmark (lengthStr) (type) Mode
Cnt Score Error Units
19:27:30 StringSerializationBenchmark.stringWrite 4 ascii thrpt
30 13453.751 ± 31.950 ops/ms
19:27:30 StringSerializationBenchmark.stringWrite 4 russian thrpt
30 8266.997 ± 15.012 ops/ms
19:27:30 StringSerializationBenchmark.stringWrite 4 chinese thrpt
30 6618.916 ± 11.011 ops/ms
19:27:30 StringSerializationBenchmark.stringWrite 128 ascii thrpt
30 754.952 ± 1.549 ops/ms
19:27:30 StringSerializationBenchmark.stringWrite 128 russian thrpt
30 420.060 ± 0.806 ops/ms
19:27:30 StringSerializationBenchmark.stringWrite 128 chinese thrpt
30 295.121 ± 0.604 ops/ms
19:27:30 StringSerializationBenchmark.stringWrite 16384 ascii thrpt
30 6.052 ± 0.014 ops/ms
19:27:30 StringSerializationBenchmark.stringWrite 16384 russian thrpt
30 3.201 ± 0.009 ops/ms
19:27:30 StringSerializationBenchmark.stringWrite 16384 chinese thrpt
30 2.268 ± 0.004 ops/ms {code}
So [~Sergey Nuyanzin] could you please spare some time to investigate on this?
> Potential regression (Jan. 13): stringWrite with Java8
> ------------------------------------------------------
>
> Key: FLINK-34148
> URL: https://issues.apache.org/jira/browse/FLINK-34148
> Project: Flink
> Issue Type: Improvement
> Components: API / Type Serialization System
> Reporter: Zakelly Lan
> Priority: Critical
> Fix For: 1.19.0
>
>
> Significant drop of performance in stringWrite with Java8 from commit
> [881062f352|https://github.com/apache/flink/commit/881062f352f8bf8c21ab7cbea95e111fd82fdf20]
> to
> [5d9d8748b6|https://github.com/apache/flink/commit/5d9d8748b64ff1a75964a5cd2857ab5061312b51]
> . It only involves strings not so long (128 or 4).
> stringWrite.128.ascii(Java8) baseline=1089.107756 current_value=754.52452
> stringWrite.128.chinese(Java8) baseline=504.244575 current_value=295.358989
> stringWrite.128.russian(Java8) baseline=655.582639 current_value=421.030188
> stringWrite.4.chinese(Java8) baseline=9598.791964 current_value=6627.929927
> stringWrite.4.russian(Java8) baseline=11070.666415 current_value=8289.95767
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
