uschindler commented on PR #12841: URL: https://github.com/apache/lucene/pull/12841#issuecomment-1871211606
You see a relic. There is no slowdown. The baseline code is exactly the same like the one The difference may come from different decisions by hotspot regarding inlining possibly because the benchmark runtime was too short. You won't the a difference in production. You are hunting ghosts. I see no reason to not backport this code. In general, I don't think the whole code here will make a big difference in production, as seen by Mike's benchmark. This is the reason why I am not happy in moving the code complexity to the indexinputs at all. So either backport it or revert it in total. These are only microbenchmarks that don't have a large impact at all. If it is twice as fast you will see a difference, but not for the code here. In production the inline decisions by hotspot will be totally different. The benchmark code only runs in isolation. -- 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
