gf2121 commented on pull request #578: URL: https://github.com/apache/lucene/pull/578#issuecomment-1005403431
Thanks @gsmiller for suggestion! > With this kind of speedup just for avoiding the null check inside the counting loop. Yes, this is indeed suprising. I thought `BrowseMonthTaxoFacets` should get a similar QPS as `BrowseMonthSSDVFacets` since SSDV is already using a `count[ord]++` and its total cardinality is just 12 (converting to global ordinal should be cheap). I also noticed that the stdDev is becoming much higher. It seems some of the JVM rounds are getting speed up but some not, I suspect some optimization in JVM was triggered from time to time but i'm not really sure what it is. I tried JDK11/16/17 and all get similar benchmark results (higher QPS & higer stdDev). > What about relaxing visibility of both values and sparseValues to protected and writing specialized counting logic for both count and countAll based on which is non-null? Nice idea, will do. -- 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
