jpountz opened a new pull request, #15160: URL: https://github.com/apache/lucene/pull/15160
With recent changes that vectorized query evaluation, Lucene could further improve CPU efficiency by bumping the block size. This reduces the likelihood of skipping blocks of postings in exchange for better CPU efficiency when processing blocks of postings. So in theory this helps the harder queries (the ones that can't skip much) and may hurt a bit the easier queries. The optimization for prefix sums that computes the sum on partially decoded data is less applicable since the introduction of the bitset encoding and with 2x bigger blocks, so it has been removed: delta-coded blocks get fully decoded before their prefix sum is computed. -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
