jpountz opened a new pull request, #13606:
URL: https://github.com/apache/lucene/pull/13606

   This iterates on #13546 to further reduce the overhead of search concurrency 
by caching whether the hit count threshold has been reached: once the threshold 
has been reached, it cannot get "un-reached" again, so we don't need to pay the 
cost of `LongAdder#get`.
   
   luceneutil on wikibigall returns the following results:
   
   ```
                               TaskQPS baseline      StdDevQPS 
my_modified_version      StdDev                Pct diff p-value
                  HighTermMonthSort     2008.33      (1.8%)     1878.71      
(0.9%)   -6.5% (  -9% -   -3%) 0.000
                         AndHighLow     1625.91      (1.9%)     1534.70      
(1.7%)   -5.6% (  -8% -   -2%) 0.000
                    CountOrHighHigh      172.89     (19.5%)      164.63     
(16.4%)   -4.8% ( -34% -   38%) 0.403
                     CountOrHighMed      339.09     (12.5%)      328.17     
(10.0%)   -3.2% ( -22% -   22%) 0.369
                       OrNotHighLow     1399.32      (2.4%)     1360.12      
(2.2%)   -2.8% (  -7% -    1%) 0.000
                             IntNRQ      979.38     (12.5%)      972.34     
(12.0%)   -0.7% ( -22% -   27%) 0.853
              HighTermDayOfYearSort      753.40      (1.8%)      748.88      
(1.3%)   -0.6% (  -3% -    2%) 0.224
                           Wildcard      356.03      (3.0%)      355.18      
(3.1%)   -0.2% (  -6% -    6%) 0.806
                            Prefix3      728.14      (3.6%)      726.78      
(4.6%)   -0.2% (  -8% -    8%) 0.887
                    CountAndHighMed      382.43      (2.0%)      382.32      
(2.6%)   -0.0% (  -4% -    4%) 0.970
                            Respell       56.41      (2.2%)       56.55      
(1.8%)    0.3% (  -3% -    4%) 0.691
                             Fuzzy2       96.54      (1.7%)       96.94      
(1.3%)    0.4% (  -2% -    3%) 0.390
                And2Terms2StopWords      358.25      (2.8%)      360.56      
(2.6%)    0.6% (  -4% -    6%) 0.455
                             Fuzzy1      104.77      (1.6%)      105.67      
(1.1%)    0.9% (  -1% -    3%) 0.052
                           PKLookup      296.23      (3.0%)      299.35      
(2.4%)    1.1% (  -4% -    6%) 0.221
                        CountPhrase        8.60      (2.3%)        8.70      
(2.6%)    1.2% (  -3% -    6%) 0.128
                             Phrase       35.20      (2.7%)       35.70      
(4.3%)    1.4% (  -5% -    8%) 0.202
                          CountTerm    11340.80      (4.2%)    11503.91      
(4.4%)    1.4% (  -6% -   10%) 0.292
                       AndStopWords       87.39      (4.3%)       88.72      
(4.3%)    1.5% (  -6% -   10%) 0.258
                   CountAndHighHigh      129.53      (3.2%)      131.56      
(2.6%)    1.6% (  -4% -    7%) 0.088
                          And3Terms      372.63      (2.5%)      380.13      
(2.3%)    2.0% (  -2% -    6%) 0.008
                        OrStopWords       90.38      (6.0%)       92.38      
(5.9%)    2.2% (  -9% -   15%) 0.242
               HighTermTitleBDVSort       67.17      (4.1%)       69.87      
(5.5%)    4.0% (  -5% -   14%) 0.009
                         OrHighRare      307.16      (9.2%)      323.41     
(10.6%)    5.3% ( -13% -   27%) 0.092
                 Or2Terms2StopWords      307.97      (3.0%)      324.57      
(2.8%)    5.4% (   0% -   11%) 0.000
                         AndHighMed      446.56      (2.6%)      473.50      
(2.8%)    6.0% (   0% -   11%) 0.000
                         TermDTSort      342.04      (2.6%)      363.80      
(2.9%)    6.4% (   0% -   12%) 0.000
                        AndHighHigh      196.61      (3.8%)      209.92      
(4.1%)    6.8% (  -1% -   15%) 0.000
                           Or3Terms      333.56      (2.7%)      361.92      
(3.2%)    8.5% (   2% -   14%) 0.000
                         OrHighHigh      196.12      (5.4%)      213.25      
(5.4%)    8.7% (  -1% -   20%) 0.000
                       OrNotHighMed      387.36      (5.5%)      426.70      
(5.1%)   10.2% (   0% -   21%) 0.000
                      OrHighNotHigh      291.03      (4.5%)      334.65      
(6.4%)   15.0% (   3% -   27%) 0.000
                      OrNotHighHigh      296.55      (4.6%)      347.48      
(7.8%)   17.2% (   4% -   31%) 0.000
                          OrHighMed      476.07      (3.0%)      558.12      
(3.0%)   17.2% (  10% -   24%) 0.000
                  HighTermTitleSort      161.76      (4.2%)      193.64      
(2.7%)   19.7% (  12% -   27%) 0.000
                          OrHighLow      811.18      (2.3%)      971.89      
(3.8%)   19.8% (  13% -   26%) 0.000
                       OrHighNotMed      399.05      (3.7%)      494.65      
(5.1%)   24.0% (  14% -   34%) 0.000
                            LowTerm      940.62      (3.6%)     1196.22      
(6.2%)   27.2% (  16% -   38%) 0.000
                            MedTerm      628.29      (3.6%)      806.56      
(5.8%)   28.4% (  18% -   39%) 0.000
                       OrHighNotLow      521.71      (4.1%)      671.57      
(5.7%)   28.7% (  18% -   40%) 0.000
                           HighTerm      473.33      (3.8%)      613.79      
(5.5%)   29.7% (  19% -   40%) 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