HUSTERGS commented on PR #14772:
URL: https://github.com/apache/lucene/pull/14772#issuecomment-2978687251
@gf2121 Thanks for your suggestion!
> I wonder if we can benchmark the following implementation to confirm the
source of speedup?
I'v done the benchmark according to your suggestion, here is the result
under identical setup
```
TaskQPS baseline StdDevQPS my_modified_version StdDev
Pct diff p-value
CombinedOrHighHigh 6.43 (3.1%) 6.33
(3.5%) -1.6% ( -7% - 5%) 0.136
CountTerm 6286.39 (5.1%) 6202.67
(5.0%) -1.3% ( -10% - 9%) 0.403
SloppyPhrase 1.48 (6.1%) 1.46
(6.1%) -1.3% ( -12% - 11%) 0.504
IntSet 391.57 (4.3%) 386.72
(3.6%) -1.2% ( -8% - 6%) 0.324
TermDTSort 191.55 (4.3%) 189.30
(4.1%) -1.2% ( -9% - 7%) 0.374
Respell 43.91 (3.4%) 43.40
(2.6%) -1.2% ( -6% - 4%) 0.224
CombinedTerm 13.53 (2.9%) 13.38
(4.6%) -1.2% ( -8% - 6%) 0.341
CombinedOrHighMed 24.60 (4.4%) 24.37
(5.7%) -0.9% ( -10% - 9%) 0.572
Prefix3 96.37 (3.2%) 95.64
(3.6%) -0.8% ( -7% - 6%) 0.481
Fuzzy2 44.58 (3.3%) 44.26
(3.9%) -0.7% ( -7% - 6%) 0.533
FilteredPrefix3 89.75 (3.5%) 89.18
(3.5%) -0.6% ( -7% - 6%) 0.567
DismaxTerm 597.24 (4.8%) 593.46
(4.1%) -0.6% ( -9% - 8%) 0.653
IntNRQ 48.68 (2.1%) 48.38
(1.9%) -0.6% ( -4% - 3%) 0.334
FilteredIntNRQ 48.35 (1.9%) 48.08
(1.9%) -0.6% ( -4% - 3%) 0.358
TermMonthSort 2333.01 (2.7%) 2321.17
(3.3%) -0.5% ( -6% - 5%) 0.597
AndHighHigh 25.77 (9.5%) 25.64
(9.0%) -0.5% ( -17% - 19%) 0.868
Fuzzy1 49.14 (3.4%) 48.95
(3.9%) -0.4% ( -7% - 7%) 0.730
Term 553.88 (7.3%) 551.72
(6.4%) -0.4% ( -13% - 14%) 0.857
OrHighRare 111.67 (7.0%) 111.30
(6.9%) -0.3% ( -13% - 14%) 0.879
FilteredOr2Terms2StopWords 63.72 (2.6%) 63.51
(2.9%) -0.3% ( -5% - 5%) 0.709
CountOrMany 5.91 (3.2%) 5.89
(3.5%) -0.3% ( -6% - 6%) 0.765
TermB1M 553.86 (7.2%) 552.13
(6.4%) -0.3% ( -12% - 14%) 0.885
DismaxOrHighHigh 43.53 (5.5%) 43.40
(5.5%) -0.3% ( -10% - 11%) 0.861
AndHighMed 64.98 (7.8%) 64.78
(8.1%) -0.3% ( -14% - 16%) 0.904
DismaxOrHighMed 62.12 (4.6%) 61.94
(5.5%) -0.3% ( -9% - 10%) 0.850
OrMany 5.32 (6.5%) 5.30
(8.1%) -0.3% ( -13% - 15%) 0.902
FilteredOrHighHigh 17.39 (1.8%) 17.35
(2.0%) -0.2% ( -3% - 3%) 0.692
TermTitleSort 62.78 (6.8%) 62.63
(6.3%) -0.2% ( -12% - 13%) 0.908
FilteredAnd3Terms 126.50 (2.5%) 126.23
(3.0%) -0.2% ( -5% - 5%) 0.813
FilteredOrStopWords 10.63 (1.7%) 10.61
(2.3%) -0.2% ( -4% - 3%) 0.764
TermB1M1P 552.80 (7.2%) 551.81
(6.4%) -0.2% ( -12% - 14%) 0.934
CountAndHighHigh 60.70 (1.4%) 60.60
(2.5%) -0.2% ( -4% - 3%) 0.797
FilteredOr3Terms 57.09 (1.9%) 57.00
(2.2%) -0.2% ( -4% - 3%) 0.809
Term10K 553.61 (7.4%) 552.84
(6.5%) -0.1% ( -13% - 14%) 0.950
OrHighHigh 24.40 (9.2%) 24.36
(8.5%) -0.1% ( -16% - 19%) 0.961
CountFilteredPhrase 11.39 (3.0%) 11.38
(3.7%) -0.1% ( -6% - 6%) 0.942
FilteredTerm 83.08 (2.2%) 83.02
(2.7%) -0.1% ( -4% - 4%) 0.922
FilteredOrHighMed 51.44 (2.4%) 51.42
(2.7%) -0.0% ( -5% - 5%) 0.957
TermDayOfYearSort 340.99 (2.7%) 340.87
(2.8%) -0.0% ( -5% - 5%) 0.967
SpanNear 3.07 (4.1%) 3.07
(4.1%) -0.0% ( -7% - 8%) 0.978
IntervalsOrdered 2.96 (3.1%) 2.96
(3.3%) -0.0% ( -6% - 6%) 0.974
AndMedOrHighHigh 20.23 (3.5%) 20.23
(3.6%) -0.0% ( -6% - 7%) 0.978
Term1M 552.05 (7.1%) 552.08
(6.4%) 0.0% ( -12% - 14%) 0.998
CombinedAndHighHigh 6.38 (3.3%) 6.38
(4.5%) 0.1% ( -7% - 8%) 0.946
CountOrHighHigh 61.87 (1.7%) 61.93
(2.8%) 0.1% ( -4% - 4%) 0.899
FilteredOrMany 4.95 (2.3%) 4.96
(1.9%) 0.1% ( -3% - 4%) 0.880
OrHighMed 84.47 (7.9%) 84.58
(8.7%) 0.1% ( -15% - 18%) 0.961
Term100 552.55 (7.2%) 553.36
(6.5%) 0.1% ( -12% - 14%) 0.946
Phrase 9.44 (3.6%) 9.46
(4.3%) 0.1% ( -7% - 8%) 0.905
CountOrHighMed 92.93 (2.0%) 93.08
(1.9%) 0.2% ( -3% - 4%) 0.790
FilteredPhrase 12.25 (2.5%) 12.27
(2.8%) 0.2% ( -4% - 5%) 0.830
CountAndHighMed 90.33 (2.7%) 90.50
(1.9%) 0.2% ( -4% - 4%) 0.794
CountFilteredOrMany 5.98 (1.7%) 5.99
(2.7%) 0.2% ( -4% - 4%) 0.787
FilteredAnd2Terms2StopWords 70.18 (4.2%) 70.32
(5.9%) 0.2% ( -9% - 10%) 0.900
CountFilteredOrHighHigh 25.00 (1.7%) 25.07
(2.1%) 0.3% ( -3% - 4%) 0.634
CountFilteredOrHighMed 29.34 (1.8%) 29.44
(2.2%) 0.3% ( -3% - 4%) 0.581
CountFilteredIntNRQ 22.12 (1.7%) 22.20
(2.0%) 0.4% ( -3% - 4%) 0.530
AndStopWords 8.83 (9.5%) 8.87
(8.0%) 0.4% ( -15% - 19%) 0.882
Wildcard 57.13 (3.9%) 57.40
(3.5%) 0.5% ( -6% - 8%) 0.693
CountPhrase 3.18 (4.2%) 3.20
(3.9%) 0.5% ( -7% - 8%) 0.693
Or3Terms 73.37 (7.4%) 73.76
(8.3%) 0.5% ( -14% - 17%) 0.834
OrStopWords 9.49 (10.6%) 9.54
(9.0%) 0.5% ( -17% - 22%) 0.864
And3Terms 81.37 (6.8%) 81.81
(7.6%) 0.5% ( -12% - 16%) 0.814
And2Terms2StopWords 67.89 (5.9%) 68.28
(8.0%) 0.6% ( -12% - 15%) 0.798
AndHighOrMedMed 17.05 (4.0%) 17.16
(3.9%) 0.6% ( -6% - 8%) 0.606
Or2Terms2StopWords 69.58 (6.0%) 70.17
(7.9%) 0.8% ( -12% - 15%) 0.706
CombinedAndHighMed 24.31 (5.5%) 24.53
(6.9%) 0.9% ( -10% - 14%) 0.659
FilteredAndHighMed 41.19 (2.8%) 42.41
(3.5%) 3.0% ( -3% - 9%) 0.003
FilteredAndStopWords 11.05 (2.2%) 11.54
(2.4%) 4.4% ( 0% - 9%) 0.000
FilteredAndHighHigh 13.89 (2.1%) 14.62
(2.1%) 5.3% ( 1% - 9%) 0.000
```
The result shows a same speedup!
--
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
