merlimat opened a new pull request, #3810:
URL: https://github.com/apache/bookkeeper/pull/3810
### Motivation
In `DigestManager.verifyDigestAndReturnData()` there are a couple of simple
improvements:
1. For the temp buffer where to store the digest, do not acquire a buffer
from the pool each time. Instead, use a thread-local buffer.
2. Instead of taking slices() at each time, extend the interface to take a
"offset, length" pair and save the allocations.
### Benchmarks:
#### Before
```
Benchmark (entrySize) Mode Cnt Score
Error Units
DigestManagerBenchmark.verifyDigest 64 thrpt 3 13533.060 ±
131.004 ops/ms
DigestManagerBenchmark.verifyDigest 1024 thrpt 3 7544.684 ±
669.817 ops/ms
DigestManagerBenchmark.verifyDigest 4086 thrpt 3 3171.914 ±
173.051 ops/ms
DigestManagerBenchmark.verifyDigest 8192 thrpt 3 1789.657 ±
33.677 ops/ms
```
Allocations: 230 bytes per operation
#### After
```
Benchmark (entrySize) Mode Cnt Score
Error Units
DigestManagerBenchmark.verifyDigest 64 thrpt 3 28108.235 ±
2319.595 ops/ms
DigestManagerBenchmark.verifyDigest 1024 thrpt 3 10245.652 ±
4175.297 ops/ms
DigestManagerBenchmark.verifyDigest 4086 thrpt 3 3527.974 ±
475.364 ops/ms
DigestManagerBenchmark.verifyDigest 8192 thrpt 3 1871.765 ±
227.182 ops/ms
```
Allocations: 64 bytes per operation
Note: these remaining 64 bytes are only due to the Java9 reflection access.
It would not be present on the JNI-based CRC32c backend (though that does not
work on Mac M1).

--
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]