jpountz commented on code in PR #14701: URL: https://github.com/apache/lucene/pull/14701#discussion_r2104624813
########## lucene/core/src/java/org/apache/lucene/search/BlockMaxConjunctionBulkScorer.java: ########## @@ -118,98 +115,36 @@ private void scoreWindow( return; } - Scorable scorer1 = this.scorer1; - if (scorers[0].getMaxScore(max - 1) == 0f) { Review Comment: This may be a contributing factor indeed. I suspect that another factor is that filtered tasks use the default impl of `nextDocsAndScores` which adds some overhead compared to what we were doing before since it first needs to copy matching docs to an array before doing work on this array. I was keen on looking into the performance of filtered tasks in a follow-up, I worry it's not going to be easy to recover performance. ########## lucene/core/src/java/org/apache/lucene/search/BlockMaxConjunctionBulkScorer.java: ########## @@ -118,98 +115,36 @@ private void scoreWindow( return; } - Scorable scorer1 = this.scorer1; - if (scorers[0].getMaxScore(max - 1) == 0f) { Review Comment: This may be a contributing factor indeed. I suspect that another factor is that filtered tasks use the default impl of `nextDocsAndScores` which adds some overhead compared to what we were doing before since it first needs to copy matching docs to an array before doing work on this array. I was keen to looking into the performance of filtered tasks in a follow-up, I worry it's not going to be easy to recover performance. -- 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