gf2121 commented on pull request #602: URL: https://github.com/apache/lucene/pull/602#issuecomment-1012836713
FYI here are some of my thoughts based on all these benchmark reports: * Rolling up loops for vint/vlong seems not bring a significant speed up (neither significant regression), IMO we should move on as loop is what vint/vlong should be since the bug will no longer occur in java 17. * The [mmap approach](https://github.com/apache/lucene/pull/592) is showing a stable speed up, and "too many codes" seems not the real reason that prevents the inline, so maybe (as @uschindler said) it is prevented by the try catch block? I'll suggest continue on the mmap approach too, but maybe we can loop the vint/vlong in `ByteBufferedIndexInput`? I'll benchmark on looped `ByteBufferedIndexInput#vint`. -- 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: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org