rmuir commented on pull request #2094: URL: https://github.com/apache/lucene-solr/pull/2094#issuecomment-734260852
I think there is plenty of opportunity to make performance regressions with this change, we should have benchmarks showing the improvement before we commit any such drastic changes to the low level interfaces? AFAIK, we already hacked away for postings-list decode to do what it needs in LE order for some vectorization reasons (LUCENE-9027). Otherwise, while its a nice goal, I am highly suspicious that the bswap is really slowing things down. Instead in practice the patch looks very complex, tons of swapping changes outside of backwards codecs (?), and even some proposals here that will definitely slow things down. So I'd like to see benchmark results before anything is committed. ---------------------------------------------------------------- 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. 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
