costin commented on PR #16356: URL: https://github.com/apache/lucene/pull/16356#issuecomment-5106734143
Appreciate the feedback. Regarding the generic wrapping approach, I opted for improving the performance of the existing filters so that users can benefit from it after the upgrade without having to change their setup/mappings. I reran the benchmark with StopFilter in the pipeline (default English/German stop sets) to see the stopword removal impact: Stopword removal helps throughput (~50% for English and ~25% for German); the cache gives 2-3x on top of it. | language | vocabSize | stopFilter | cacheSize=0 | cacheSize=1024 | speedup | |----------|----------:|------------|---------------:|----------------:|--------:| | English | 500 | true | 409.0 ± 5.4 | 861.1 ± 4.9 | 2.11x | | English | 500 | false | 278.7 ± 2.2 | 758.5 ± 1.6 | 2.72x | | English | 5,000 | true | 403.7 ± 2.2 | 820.7 ± 5.9 | 2.03x | | English | 5,000 | false | 280.3 ± 3.1 | 673.7 ± 7.0 | 2.40x | | English | 50,000 | true | 389.3 ± 2.1 | 771.2 ± 2.3 | 1.98x | | English | 50,000 | false | 269.0 ± 10.0 | 649.0 ± 1.6 | 2.41x | | German | 500 | true | 242.8 ± 4.1 | 690.8 ± 5.9 | 2.84x | | German | 500 | false | 192.6 ± 3.0 | 553.0 ± 3.6 | 2.87x | | German | 5,000 | true | 228.5 ± 3.1 | 624.3 ± 8.1 | 2.73x | | German | 5,000 | false | 184.6 ± 3.5 | 494.6 ± 5.8 | 2.68x | | German | 50,000 | true | 232.2 ± 3.1 | 593.7 ± 3.3 | 2.56x | | German | 50,000 | false | 184.2 ± 4.0 | 464.4 ± 6.1 | 2.52x | (same hardware and JDK as before) -- 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]
