easyice commented on PR #12841:
URL: https://github.com/apache/lucene/pull/12841#issuecomment-1852253188
Even if i used the copy code approach(avoid to using lambda, for test
purpose), it was only 15%-20% faster.
like this:
```
@Override
public void readGroupVInts(long[] dst, int limit) throws IOException {
int i;
for (i = 0; i <= limit - 4; i += 4) {
readGroupVInt(dst, i);
}
for (; i < limit; ++i) {
dst[i] = readVInt();
}
}
private void readGroupVInt(long[] dst, int offset) throws IOException {
final ByteBuffer buffer = curBuf;
if (buffer.remaining() < GroupVIntUtil.MAX_LENGTH_PER_GROUP) {
GroupVIntUtil.readGroupVInt(this, dst, offset);
return;
}
final int flag = readByte() & 0xFF;
...
dst[offset] = guard.getInt(buffer, curPosition) &
GroupVIntUtil.MASKS[n1Minus1];
...
buffer.position(curPosition);
}
```
copy code approach output:
```
Benchmark (size)
Mode Cnt Score Error Units
GroupVIntBenchmark.benchMMapDirectoryInputs_readGroupVInt 64
thrpt 5 8.991 ± 1.569 ops/us
GroupVIntBenchmark.benchMMapDirectoryInputs_readGroupVIntBaseline 64
thrpt 5 7.387 ± 0.268 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]
