rmuir commented on issue #14042: URL: https://github.com/apache/lucene/issues/14042#issuecomment-2537347402
> I've yet to do any real debugging, but my sense here is that the aggressive unrolling in that we do (4x with float dot), is hurting us when we inline. I'll try to reproduce on my Intel box, this might not be AMD specific (might just hurt more there). I wanna run luceneutil benchmarks rather than the micro-benchmarks. what else are we to do though? cpus have multiple fma units, jvm won't unroll as it will change results of floating point. -- 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
