fulmicoton commented on issue #12358: URL: https://github.com/apache/lucene/issues/12358#issuecomment-1585647850
> Of course a real world benchmark should also measure throughput by hammering a server with hundreds of parallel queries (many more than there are CPU cores) to saturate all CPU cores. The goal of the original benchmark is to identify performance headrooms in tantivy. That's why it measures different kind of queries, does not deal with multithreading, works over a single segment, and compares "min" and not "mean". "min" removes a lot of the noise and is for this reason a bit generous for Java... but it is much better to give us the right insights. BTW I suspect changing the GC won't have any effect on the min statistics, but it does not hurt to try. I'd love to see a fair bench of multithreaded search throughput / indexing of course. Making it fair is however much trickier. -- 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
