gsmiller commented on pull request #597:
URL: https://github.com/apache/lucene/pull/597#issuecomment-1011267849
Here are results of just reverting LUCENE-10245. This doesn't look like the
cause of the regression to me. None of the results are significant, and if
anything, the tasks associated with the regression appear to trend towards
further regression (but taken with a grain of salt given the p-values):
```
TaskQPS baseline StdDevQPS candidate
StdDev Pct diff p-value
Prefix3 183.53 (10.7%) 173.75
(13.0%) -5.3% ( -26% - 20%) 0.157
BrowseDayOfYearTaxoFacets 30.40 (28.9%) 29.25
(30.5%) -3.8% ( -49% - 78%) 0.688
BrowseDateTaxoFacets 29.99 (28.7%) 28.87
(30.3%) -3.8% ( -48% - 77%) 0.687
BrowseRandomLabelTaxoFacets 23.17 (28.6%) 22.42
(30.0%) -3.2% ( -48% - 77%) 0.727
HighTermDayOfYearSort 118.53 (15.4%) 115.81
(16.6%) -2.3% ( -29% - 35%) 0.651
Wildcard 94.18 (3.3%) 92.49
(4.7%) -1.8% ( -9% - 6%) 0.161
HighTermTitleBDVSort 23.85 (29.3%) 23.46
(23.6%) -1.7% ( -42% - 72%) 0.843
TermDTSort 116.28 (15.5%) 114.63
(16.7%) -1.4% ( -29% - 36%) 0.780
OrHighMedDayTaxoFacets 12.28 (4.5%) 12.11
(5.2%) -1.3% ( -10% - 8%) 0.387
MedTerm 1707.68 (3.2%) 1685.74
(2.7%) -1.3% ( -6% - 4%) 0.170
LowTerm 2003.28 (3.4%) 1985.24
(3.5%) -0.9% ( -7% - 6%) 0.411
Fuzzy2 24.44 (2.6%) 24.22
(2.7%) -0.9% ( -6% - 4%) 0.283
AndHighHigh 41.84 (4.6%) 41.50
(4.6%) -0.8% ( -9% - 8%) 0.587
AndHighLow 1251.13 (1.8%) 1241.38
(2.3%) -0.8% ( -4% - 3%) 0.236
HighTermMonthSort 157.34 (16.4%) 156.13
(16.3%) -0.8% ( -28% - 38%) 0.882
Fuzzy1 78.31 (2.2%) 77.71
(2.4%) -0.8% ( -5% - 3%) 0.297
AndHighMed 91.13 (4.3%) 90.43
(4.3%) -0.8% ( -9% - 8%) 0.578
MedPhrase 30.97 (3.1%) 30.74
(4.0%) -0.7% ( -7% - 6%) 0.514
OrNotHighHigh 892.97 (3.1%) 887.10
(3.6%) -0.7% ( -7% - 6%) 0.537
LowPhrase 32.18 (3.3%) 31.97
(4.1%) -0.6% ( -7% - 7%) 0.582
Respell 56.57 (2.3%) 56.25
(2.4%) -0.6% ( -5% - 4%) 0.455
BrowseRandomLabelSSDVFacets 9.46 (2.7%) 9.42
(2.4%) -0.5% ( -5% - 4%) 0.536
HighPhrase 188.03 (3.0%) 187.17
(3.3%) -0.5% ( -6% - 5%) 0.644
OrHighNotHigh 822.83 (3.4%) 819.58
(3.6%) -0.4% ( -7% - 6%) 0.721
OrHighNotLow 1104.21 (3.3%) 1099.89
(3.1%) -0.4% ( -6% - 6%) 0.701
OrHighNotMed 992.14 (3.8%) 988.68
(3.5%) -0.3% ( -7% - 7%) 0.765
IntNRQ 94.40 (2.2%) 94.13
(2.5%) -0.3% ( -4% - 4%) 0.703
OrNotHighLow 995.88 (1.9%) 993.32
(2.2%) -0.3% ( -4% - 3%) 0.696
PKLookup 171.16 (3.2%) 170.92
(3.7%) -0.1% ( -6% - 6%) 0.899
AndHighHighDayTaxoFacets 28.82 (2.8%) 28.79
(2.7%) -0.1% ( -5% - 5%) 0.910
AndHighMedDayTaxoFacets 110.31 (2.0%) 110.26
(1.8%) -0.0% ( -3% - 3%) 0.935
LowSloppyPhrase 71.16 (2.1%) 71.17
(2.6%) 0.0% ( -4% - 4%) 0.992
MedTermDayTaxoFacets 47.28 (4.5%) 47.33
(3.1%) 0.1% ( -7% - 8%) 0.941
OrNotHighMed 1418.54 (2.2%) 1420.09
(3.2%) 0.1% ( -5% - 5%) 0.901
MedSpanNear 30.86 (1.7%) 30.95
(1.6%) 0.3% ( -2% - 3%) 0.568
LowSpanNear 31.93 (1.5%) 32.04
(1.4%) 0.3% ( -2% - 3%) 0.492
HighTerm 1439.92 (3.4%) 1444.80
(3.1%) 0.3% ( -6% - 7%) 0.744
OrHighLow 319.28 (2.1%) 320.62
(3.0%) 0.4% ( -4% - 5%) 0.612
OrHighMed 123.42 (3.3%) 123.96
(3.7%) 0.4% ( -6% - 7%) 0.688
BrowseDayOfYearSSDVFacets 12.89 (15.0%) 12.95
(15.9%) 0.5% ( -26% - 36%) 0.925
HighSpanNear 21.17 (2.6%) 21.27
(2.2%) 0.5% ( -4% - 5%) 0.529
HighSloppyPhrase 40.75 (3.3%) 40.99
(3.4%) 0.6% ( -5% - 7%) 0.585
MedSloppyPhrase 24.95 (3.0%) 25.10
(3.1%) 0.6% ( -5% - 6%) 0.523
OrHighHigh 57.63 (3.3%) 58.03
(3.5%) 0.7% ( -5% - 7%) 0.517
MedIntervalsOrdered 11.61 (4.4%) 11.77
(3.0%) 1.3% ( -5% - 9%) 0.260
HighIntervalsOrdered 8.61 (6.2%) 8.79
(4.4%) 2.0% ( -7% - 13%) 0.228
LowIntervalsOrdered 176.87 (7.6%) 181.67
(5.5%) 2.7% ( -9% - 17%) 0.195
BrowseMonthSSDVFacets 14.19 (20.6%) 14.72
(25.3%) 3.7% ( -34% - 62%) 0.611
BrowseMonthTaxoFacets 28.48 (28.8%) 29.94
(22.1%) 5.1% ( -35% - 78%) 0.526
```
--
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]
