jpountz commented on PR #14133:
URL: https://github.com/apache/lucene/pull/14133#issuecomment-2586551681

   Opening as a draft for now because I would like to change the way how 
deleted docs are applied with the `#loadIntoBitSet` API. As things are today, a 
single deleted doc in a segment would completely cancel the speedup.
   
   Here is what `luceneutil` reports on `wikibigall`:
   
   ```
                               TaskQPS baseline      StdDevQPS 
my_modified_version      StdDev                Pct diff p-value
                      TermTitleSort      153.45      (2.6%)      146.17      
(2.1%)   -4.7% (  -9% -    0%) 0.000
                            Prefix3      140.30      (4.3%)      134.27      
(3.1%)   -4.3% ( -11% -    3%) 0.008
                        OrStopWords       34.73      (7.9%)       33.26      
(8.9%)   -4.2% ( -19% -   13%) 0.239
                    FilteredPrefix3      133.93      (4.2%)      128.54      
(3.0%)   -4.0% ( -10% -    3%) 0.009
                               Term      487.39      (2.8%)      470.53      
(4.9%)   -3.5% ( -10% -    4%) 0.042
                         DismaxTerm      584.95      (2.1%)      568.87      
(3.8%)   -2.7% (  -8% -    3%) 0.037
                           Or3Terms      172.20      (4.7%)      167.58      
(5.1%)   -2.7% ( -11% -    7%) 0.203
                         OrHighHigh       54.62      (6.3%)       53.22      
(4.4%)   -2.6% ( -12% -    8%) 0.270
                           Wildcard       78.76      (3.6%)       77.01      
(3.3%)   -2.2% (  -8% -    4%) 0.131
                    AndHighOrMedMed       44.90      (1.0%)       43.91      
(1.2%)   -2.2% (  -4% -    0%) 0.000
                             OrMany       19.43      (2.7%)       19.03      
(4.4%)   -2.1% (  -8% -    5%) 0.184
                         TermDTSort      286.84      (7.8%)      281.19      
(5.9%)   -2.0% ( -14% -   12%) 0.505
                        AndHighHigh       44.74      (1.5%)       43.87      
(2.4%)   -2.0% (  -5% -    1%) 0.022
                         OrHighRare      278.82      (6.4%)      273.59      
(7.8%)   -1.9% ( -15% -   13%) 0.537
                             Fuzzy1       81.43      (2.6%)       80.06      
(2.1%)   -1.7% (  -6% -    3%) 0.101
                       CombinedTerm       31.73      (2.2%)       31.20      
(2.5%)   -1.7% (  -6% -    3%) 0.105
                          And3Terms      173.65      (3.4%)      170.79      
(3.9%)   -1.6% (  -8% -    5%) 0.289
                     FilteredOrMany       16.75      (1.4%)       16.48      
(2.8%)   -1.6% (  -5% -    2%) 0.098
                      TermMonthSort     3380.63      (3.0%)     3329.65      
(2.1%)   -1.5% (  -6% -    3%) 0.175
                 CombinedOrHighHigh       19.05      (1.8%)       18.77      
(1.5%)   -1.5% (  -4% -    1%) 0.036
                       AndStopWords       31.57      (4.1%)       31.11      
(6.6%)   -1.5% ( -11% -    9%) 0.532
                             Fuzzy2       76.49      (2.2%)       75.47      
(1.8%)   -1.3% (  -5% -    2%) 0.117
                         AndHighMed      128.95      (1.2%)      127.28      
(2.9%)   -1.3% (  -5% -    2%) 0.173
                 Or2Terms2StopWords      162.86      (4.8%)      161.03      
(5.2%)   -1.1% ( -10% -    9%) 0.601
                        CountPhrase        4.18      (1.6%)        4.14      
(7.7%)   -1.0% ( -10% -    8%) 0.684
                           PKLookup      278.86      (2.4%)      276.45      
(1.5%)   -0.9% (  -4% -    3%) 0.308
                   DismaxOrHighHigh      119.43      (4.4%)      118.55      
(4.0%)   -0.7% (  -8% -    8%) 0.682
                  FilteredAnd3Terms      192.64      (2.2%)      191.75      
(2.1%)   -0.5% (  -4% -    3%) 0.619
                And2Terms2StopWords      161.87      (3.3%)      161.43      
(3.4%)   -0.3% (  -6% -    6%) 0.848
                   FilteredOr3Terms      164.22      (1.5%)      163.83      
(1.1%)   -0.2% (  -2% -    2%) 0.673
                CombinedAndHighHigh       15.26      (1.9%)       15.23      
(1.9%)   -0.2% (  -4% -    3%) 0.774
                     FilteredIntNRQ      110.08     (12.4%)      109.86     
(13.6%)   -0.2% ( -23% -   29%) 0.971
                  CombinedOrHighMed       71.97      (1.9%)       71.85      
(1.7%)   -0.2% (  -3% -    3%) 0.827
                          CountTerm     9414.23      (5.4%)     9409.07      
(4.3%)   -0.1% (  -9% -   10%) 0.979
                             IntNRQ      110.97     (11.7%)      111.34     
(13.8%)    0.3% ( -22% -   29%) 0.951
                 FilteredAndHighMed      128.34      (2.7%)      129.86      
(2.8%)    1.2% (  -4% -    6%) 0.312
                 CombinedAndHighMed       55.25      (1.8%)       56.01      
(2.0%)    1.4% (  -2% -    5%) 0.086
                  FilteredOrHighMed      152.30      (1.4%)      154.67      
(1.3%)    1.6% (  -1% -    4%) 0.006
         FilteredOr2Terms2StopWords      146.27      (1.8%)      148.57      
(1.2%)    1.6% (  -1% -    4%) 0.016
                          OrHighMed      196.36      (5.1%)      199.53      
(3.6%)    1.6% (  -6% -   10%) 0.389
                    DismaxOrHighMed      170.18      (3.3%)      173.26      
(2.6%)    1.8% (  -4% -    8%) 0.158
                             Phrase       14.72      (5.4%)       15.09      
(5.4%)    2.5% (  -7% -   14%) 0.278
        FilteredAnd2Terms2StopWords      194.56      (1.6%)      200.16      
(1.9%)    2.9% (   0% -    6%) 0.000
                       FilteredTerm      154.40      (1.7%)      159.20      
(1.7%)    3.1% (   0% -    6%) 0.000
                  TermDayOfYearSort      628.53      (4.8%)      657.48      
(4.5%)    4.6% (  -4% -   14%) 0.021
                 FilteredOrHighHigh       64.07      (1.8%)       67.55      
(2.1%)    5.4% (   1% -    9%) 0.000
                CountFilteredPhrase       24.42      (1.8%)       26.13      
(2.8%)    7.0% (   2% -   11%) 0.000
                FilteredOrStopWords       43.16      (1.9%)       46.70      
(2.5%)    8.2% (   3% -   12%) 0.000
                   AndMedOrHighHigh       60.21      (1.7%)       66.11      
(2.1%)    9.8% (   5% -   13%) 0.000
                FilteredAndHighHigh       61.91      (1.7%)       68.38      
(2.2%)   10.5% (   6% -   14%) 0.000
                     FilteredPhrase       29.38      (1.1%)       32.96      
(2.0%)   12.2% (   9% -   15%) 0.000
               FilteredAndStopWords       47.21      (1.6%)       54.89      
(2.5%)   16.3% (  12% -   20%) 0.000
                    CountAndHighMed      238.97      (2.3%)      294.13      
(2.8%)   23.1% (  17% -   28%) 0.000
             CountFilteredOrHighMed       88.21      (0.9%)      116.76      
(0.8%)   32.4% (  30% -   34%) 0.000
            CountFilteredOrHighHigh       71.30      (1.2%)      105.22      
(1.2%)   47.6% (  44% -   50%) 0.000
                     CountOrHighMed      189.16      (2.3%)      343.97      
(4.2%)   81.8% (  73% -   90%) 0.000
                CountFilteredOrMany       11.05      (2.3%)       24.68      
(4.3%)  123.2% ( 113% -  132%) 0.000
                   CountAndHighHigh      132.06      (2.4%)      295.79      
(5.1%)  124.0% ( 113% -  134%) 0.000
                    CountOrHighHigh      123.04      (2.4%)      279.03      
(5.1%)  126.8% ( 116% -  137%) 0.000
                        CountOrMany       11.67      (1.4%)       28.06      
(5.9%)  140.4% ( 131% -  149%) 0.000
   ```


-- 
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

Reply via email to