msokolov commented on PR #15065:
URL: https://github.com/apache/lucene/pull/15065#issuecomment-3188424008
I ran
java -jar
lucene/benchmark-jmh/build/benchmarks/lucene-benchmark-jmh-11.0.0-SNAPSHOT.jar
'.*VInt.*'
on a Graviton 2 host, and the results are below. My summary is: (1) some
benchmarks aren't available on main so we have no basis for comparison (how did
you run those!), (2) for those in both, all were the same except
`GroupVIntBenchmark.benchByteArrayDataInput_readVInt`: 3.367 -> 5.153; a
~53% improvement
`GroupVIntBenchmark.benchMMapDirectoryInputs_readGroupVInt`: 3.4 -> 4.851; a
~43% improvement
LGTM there although quite different from MacBook. I'm also not less
interested in Graviton 2 than Graviton 3, 4, and Intel, it's just what I
happened to have handy
##baseline
```
Benchmark (size)
Mode Cnt Score Error Units
GroupVIntBenchmark.benchByteArrayDataInput_readGroupVInt 64
thrpt 5 7.102 ± 0.009 ops/us
GroupVIntBenchmark.benchByteArrayDataInput_readVInt 64
thrpt 5 3.367 ± 0.001 ops/us
GroupVIntBenchmark.benchByteBuffersIndexInput_readGroupVInt 64
thrpt 5 0.656 ± 0.001 ops/us
GroupVIntBenchmark.benchByteBuffersIndexInput_readGroupVIntBaseline 64
thrpt 5 0.654 ± 0.001 ops/us
GroupVIntBenchmark.benchMMapDirectoryInputs_readGroupVInt 64
thrpt 5 3.368 ± 0.004 ops/us
GroupVIntBenchmark.benchMMapDirectoryInputs_readGroupVIntBaseline 64
thrpt 5 3.372 ± 0.011 ops/us
GroupVIntBenchmark.benchMMapDirectoryInputs_readVInt 64
thrpt 5 3.400 ± 0.001 ops/us
GroupVIntBenchmark.benchNIOFSDirectoryInputs_readGroupVInt 64
thrpt 5 3.800 ± 0.001 ops/us
GroupVIntBenchmark.benchNIOFSDirectoryInputs_readGroupVIntBaseline 64
thrpt 5 3.835 ± 0.001 ops/us
GroupVIntBenchmark.bench_writeGroupVInt 64
thrpt 5 1.812 ± 0.003 ops/us
```
## candidate
```
Benchmark (size)
Mode Cnt Score Error Units
GroupVIntBenchmark.benchByteArrayDataInput_readGroupVInt 64
thrpt 5 7.102 ± 0.002 ops/us
GroupVIntBenchmark.benchByteArrayDataInput_readVInt 64
thrpt 5 5.153 ± 0.021 ops/us
GroupVIntBenchmark.benchByteBuffersIndexInput_readGroupVInt 64
thrpt 5 0.658 ± 0.001 ops/us
GroupVIntBenchmark.benchByteBuffersIndexInput_readGroupVIntBaseline 64
thrpt 5 0.656 ± 0.001 ops/us
GroupVIntBenchmark.benchMMapDirectoryInputs_readGroupVInt 64
thrpt 5 3.368 ± 0.004 ops/us
GroupVIntBenchmark.benchMMapDirectoryInputs_readGroupVIntBaseline 64
thrpt 5 3.373 ± 0.004 ops/us
GroupVIntBenchmark.benchMMapDirectoryInputs_readVInt 64
thrpt 5 4.851 ± 0.059 ops/us
GroupVIntBenchmark.benchNIOFSDirectoryInputs_readGroupVInt 64
thrpt 5 3.836 ± 0.001 ops/us
GroupVIntBenchmark.benchNIOFSDirectoryInputs_readGroupVIntBaseline 64
thrpt 5 3.835 ± 0.001 ops/us
GroupVIntBenchmark.bench_writeGroupVInt 64
thrpt 5 1.814 ± 0.006 ops/us
VIntBenchmark.benchByteArrayDataInput_readVInt 64
thrpt 5 5.193 ± 0.006 ops/us
VIntBenchmark.benchByteArrayDataInput_readVLong 64
thrpt 5 2.753 ± 0.004 ops/us
VIntBenchmark.benchMMapDirectoryInputs_readVInt 64
thrpt 5 4.839 ± 0.015 ops/us
VIntBenchmark.benchMMapDirectoryInputs_readVLong 64
thrpt 5 2.536 ± 0.001 ops/us
```
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
