zhaih commented on PR #12660: URL: https://github.com/apache/lucene/pull/12660#issuecomment-1763695626
Thanks @msokolov, I'll take a look soon. > so ideally we'd compare total time to merge using single-threaded vs using this. The most fair comparison here I posted might be the force merge time for 100k docs, which is 20012ms vs 97579ms. Since there should be no merging happening during normal indexing due to index size (around 300mb and per segment 50mb buffer), all the conditions are the same, but still we're counting the flush time and preparation time during the comparison. Because the lock are only used in several places, I'm wondering whether we could simply just accumulate the lock waitting time using `System.nanoTime()` to get a feel of how much contention I have. But in general a nearly 5x speed up over 8 thread is not bad I feel. -- 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
