jpountz commented on PR #14910:
URL: https://github.com/apache/lucene/pull/14910#issuecomment-3059176116
This is very cool and the speedup makes sense to me. When dynamic pruning is
enabled, only queries whose leading clauses are dense benefit significantly
from this speedup (`OrStopWords` and `AndStopWords`). But if you evaluated
exhaustive evaluation, I'm sure we'd be seeing a bigger speedup on all
disjunctive queries that have one dense postings list or more.
Like for #14896, I'd like to split this PR in two: one where we merge your
scalar improvements, and then this one where we add support for vectorization.
By the way, we may want to look into other approaches for the scalar case.
Since we only use bit sets in postings when many bits would be set, a linear
scan should perform quite efficiently? (`foreach (bit in 0..n) { if
bitSet.get(bit) out.append(bit); }`) I imagine that you used a micro benchmark
to come up with your manual unrolling, let's include this micro benchmark in
the PR?
--
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
