shubhamvishu commented on code in PR #12868:
URL: https://github.com/apache/lucene/pull/12868#discussion_r1415528939
##########
lucene/codecs/src/java/org/apache/lucene/codecs/bloom/FuzzySet.java:
##########
@@ -150,9 +150,10 @@ private FuzzySet(FixedBitSet filter, int bloomSize, int
hashCount) {
* @return NO or MAYBE
*/
public ContainsResult contains(BytesRef value) {
- long hash = hashFunction.hash(value);
- int msb = (int) (hash >>> Integer.SIZE);
- int lsb = (int) hash;
+ long[] hash = hashFunction.hash128(value);
+
+ int msb = ((int) hash[0] >>> Integer.SIZE) >>> 1 + ((int) hash[1] >>>
Integer.SIZE) >>> 1;
+ int lsb = ((int) hash[0]) >>> 1 + ((int) hash[1]) >>> 1;
Review Comment:
Ok I tried using simple `(hash[0] >>> Integer.SIZE)` and `(int) hash[0]` but
I still see the same 2.6% regression. I think the current way seem to produce
better positive hash value and more random than directly using the 2 long
values which is why we are seeing `PKLookup` performance getting improved by
7-9%.
--
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]
