original-brownbear commented on PR #13544:
URL: https://github.com/apache/lucene/pull/13544#issuecomment-2212636552
Thanks Adrien!
> do you have a benchmark / profiler output that shows the improvement?
The nightlies show this:
```
0.49% 4090M
java.lang.invoke.DirectMethodHandle#allocateInstance()
at
java.lang.invoke.DirectMethodHandle$Holder#newInvokeSpecial()
at
java.lang.invoke.Invokers$Holder#linkToTargetMethod()
at java.util.Comparator#comparingDouble()
at
org.apache.lucene.search.MaxScoreBulkScorer#partitionScorers()
at
org.apache.lucene.search.MaxScoreBulkScorer#score()
at org.apache.lucene.search.BulkScorer#score()
at
org.apache.lucene.search.IndexSearcher#search()
```
This change makes it go away :) So it's essentially saving 0.5% of all
allocations here. I couldn't reproduce a big/signifcant QPS change as a result
since we're not CPU bound when running the benchmark (this might change with
the `AtomicLong` change in the other PR ;)), but I think saving those GBs from
getting allocated in the first place is enough reason to fix this kind of spot
:)
--
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
