mikemccand commented on issue #15662:
URL: https://github.com/apache/lucene/issues/15662#issuecomment-3846861971
Hi Team -- sorry, I messed up the one-off benchy last night (kicked it off
"directly" but forgot to disable the `systemd` timer for 6 PM, and also I left
an emacs lock file in a bad place and broke that one (6 PM) too!
`InsufficientCaffeineException`!
I have just now tried, again, to kick off a one-off, just the direct
("tasks") benchmark (skipping stored fields stress test, KNN nightly benchy,
NAD (high cardinality facets), etc.). And this time I stopped and disabled the
timer...
Digging a bit more into this weird slowdown...
Curiously, [stored fields benchy saw no
slowdown](https://benchmarks.mikemccandless.com/stored_fields_benchmarks.html)
from 1/22 -> 1/29, and maybe a small speedup in `BEST_COMPRESSION` indexing
time (354 -> 346 sec).
[NAD high cardinality facets also looks
unaffected](https://benchmarks.mikemccandless.com/nad_facet_benchmarks.html)
1/22 -> 1/29.
[Geospatial benchmarks (OpenStreetMaps) were
affected](https://benchmarks.mikemccandless.com/geobench.html).
[Sparse benchy (NYC green/yellow taxis corpus) also looks
unaffected](https://benchmarks.mikemccandless.com/sparseResults.html), but
[maybe `TermQuery` got
faster](https://benchmarks.mikemccandless.com/sparseResults.html#search_qps)!?
Hmm we somehow [stopped running the `CombinedFieldsQuery`
benchy](https://benchmarks.mikemccandless.com/CombinedHighMed.html) (data ended
Nov 2024!) -- I'll open a luceneutil issue.
Holy smokes look [how much faster `precommit` and `gradle -p lucene test`
became recently](https://benchmarks.mikemccandless.com/antcleantest.html)!
[`CheckIndex` time seems to get slower for the past
year](https://benchmarks.mikemccandless.com/checkIndexTime.html)? (That's the
standard "search tasks" benchy's search index). Yet [sparse (NYC taxis)
`CheckIndex` time is now
crazy-fast](https://benchmarks.mikemccandless.com/sparseResults.html#check_index_time).
OK the first one is with `-threadCount 16 -level 2` (try extra hard?), but
the 2nd one is default (just checksum integrity?).
--
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]
