vsop-479 commented on PR #16245:
URL: https://github.com/apache/lucene/pull/16245#issuecomment-4967500313

   @mikemccand I ran luceneutil on `wikimediumall`. The results are mixed: some 
tasks improved, some tasks regressed.
   
   Note: the two indexes did not end up with the same segmentlayout. The 
baseline has 23 segments(`Reader has 15 slices, from 23 segments`), while the 
candidate has 15 segments(`Reader has 11 slices, from 15 segments`).
   
   # search perf
   ```
                   TaskQPS baseline      StdDevQPS my_modified_version      
StdDev                Pct diff p-value
                LowIntervalsOrdered      370.76      (7.8%)      226.93      
(2.8%)  -38.8% ( -45% -  -30%) 0.000
                         TermDTSort      283.03      (8.7%)      212.50      
(2.8%)  -24.9% ( -33% -  -14%) 0.000
                MedIntervalsOrdered       84.69      (9.1%)       65.34      
(5.7%)  -22.8% ( -34% -   -8%) 0.000
                           BM25MSM2       32.55      (6.0%)       25.90      
(4.8%)  -20.4% ( -29% -  -10%) 0.000
               HighIntervalsOrdered       46.12      (6.8%)       36.94      
(5.9%)  -19.9% ( -30% -   -7%) 0.000
                    LowSloppyPhrase       37.97      (5.2%)       30.43      
(4.5%)  -19.9% ( -28% -  -10%) 0.000
                        LowSpanNear      341.24      (7.7%)      275.33      
(6.4%)  -19.3% ( -31% -   -5%) 0.000
              HighTermDayOfYearSort      263.26      (8.3%)      212.94      
(4.6%)  -19.1% ( -29% -   -6%) 0.000
               HighTermTitleBDVSort       37.46      (4.2%)       30.65      
(7.1%)  -18.2% ( -28% -   -7%) 0.000
                             IntNRQ      136.34      (7.6%)      112.03      
(6.2%)  -17.8% ( -29% -   -4%) 0.000
                          MedPhrase      326.85      (6.8%)      277.01      
(8.4%)  -15.2% ( -28% -    0%) 0.000
                      OrHighNotHigh      877.90     (10.9%)      760.14     
(14.3%)  -13.4% ( -34% -   13%) 0.001
                        MedSpanNear      161.20      (6.4%)      139.76      
(6.6%)  -13.3% ( -24% -    0%) 0.000
                    MedSloppyPhrase      114.43      (6.4%)       99.46      
(7.4%)  -13.1% ( -25% -    0%) 0.000
                       HighSpanNear       23.63      (3.7%)       20.57      
(4.1%)  -12.9% ( -20% -   -5%) 0.000
                         AndHighMed      503.88     (13.7%)      444.40      
(8.5%)  -11.8% ( -29% -   12%) 0.001
                          OrHighMed      669.35     (10.6%)      592.70      
(6.4%)  -11.5% ( -25% -    6%) 0.000
                         OrHighHigh      271.73     (13.1%)      240.62     
(11.3%)  -11.4% ( -31% -   14%) 0.004
           AndHighHighDayTaxoFacets        8.85      (4.9%)        7.90      
(3.8%)  -10.8% ( -18% -   -2%) 0.000
                        AndHighHigh      226.98     (11.7%)      206.78     
(10.9%)   -8.9% ( -28% -   15%) 0.015
                            Prefix3     1199.87     (10.0%)     1106.83     
(11.0%)   -7.8% ( -26% -   14%) 0.023
            AndHighMedDayTaxoFacets      186.03      (4.7%)      172.02      
(6.7%)   -7.5% ( -18% -    3%) 0.000
             OrHighMedDayTaxoFacets       14.97      (4.1%)       14.11      
(4.1%)   -5.8% ( -13% -    2%) 0.000
                         HighPhrase       78.30      (5.7%)       74.79      
(6.4%)   -4.5% ( -15% -    8%) 0.023
                   HighSloppyPhrase       16.26      (4.3%)       15.69      
(2.9%)   -3.5% ( -10% -    3%) 0.003
                       OrHighNotLow     1640.86      (9.2%)     1585.33      
(6.4%)   -3.4% ( -17% -   13%) 0.188
               MedTermDayTaxoFacets       48.91      (4.6%)       47.46      
(4.0%)   -3.0% ( -11% -    5%) 0.035
                       OrHighNotMed      831.80     (14.9%)      808.40      
(8.9%)   -2.8% ( -23% -   24%) 0.479
        BrowseRandomLabelSSDVFacets        5.79      (8.9%)        5.65     
(11.3%)   -2.4% ( -20% -   19%) 0.461
          BrowseDayOfYearSSDVFacets        7.21     (11.1%)        7.05     
(12.7%)   -2.2% ( -23% -   24%) 0.570
                  HighTermTitleSort      130.27      (9.6%)      127.68      
(6.6%)   -2.0% ( -16% -   15%) 0.455
                       OrNotHighLow     1669.61      (8.4%)     1638.80     
(11.2%)   -1.8% ( -19% -   19%) 0.565
                          LowPhrase      581.10      (8.7%)      577.82     
(10.9%)   -0.6% ( -18% -   20%) 0.860
                          OrHighLow     1452.82      (8.1%)     1461.18      
(9.0%)    0.6% ( -15% -   19%) 0.836
                             IntSet      892.43     (16.9%)      903.18     
(13.4%)    1.2% ( -24% -   37%) 0.808
              BrowseMonthSSDVFacets        7.35     (15.3%)        7.53     
(23.1%)    2.5% ( -31% -   48%) 0.698
               BrowseDateSSDVFacets        1.34     (10.6%)        1.39     
(11.3%)    4.1% ( -16% -   29%) 0.252
                         AndHighLow     1576.04     (12.0%)     1643.83     
(14.1%)    4.3% ( -19% -   34%) 0.312
                             Fuzzy1      127.34      (4.5%)      135.47      
(4.7%)    6.4% (  -2% -   16%) 0.000
                            MedTerm     1431.35     (10.7%)     1523.34     
(11.3%)    6.4% ( -14% -   31%) 0.072
                     AndMissingHigh     4663.20     (12.2%)     5044.04     
(12.3%)    8.2% ( -14% -   37%) 0.040
                              range     4585.52     (10.5%)     4986.56     
(12.1%)    8.7% ( -12% -   34%) 0.017
                           HighTerm     1075.60     (11.3%)     1182.52     
(13.6%)    9.9% ( -13% -   39%) 0.014
                       OrNotHighMed      568.44     (10.1%)      635.47     
(20.7%)   11.8% ( -17% -   47%) 0.025
                           PKLookup      417.41      (3.4%)      470.58      
(5.0%)   12.7% (   4% -   21%) 0.000
                          ConstMSM2     1673.63     (13.9%)     1907.07     
(10.3%)   13.9% (  -9% -   44%) 0.000
                           Wildcard       20.58      (3.3%)       23.75      
(5.3%)   15.4% (   6% -   24%) 0.000
                            LowTerm     1875.89     (20.2%)     2208.92     
(18.8%)   17.8% ( -17% -   71%) 0.005
                             Fuzzy2       92.75      (3.4%)      109.89      
(5.3%)   18.5% (   9% -   28%) 0.000
                  HighTermMonthSort     1684.36      (9.1%)     1997.24     
(10.8%)   18.6% (  -1% -   42%) 0.000
                            Respell       93.80      (3.0%)      112.09      
(4.1%)   19.5% (  12% -   27%) 0.000
                      OrNotHighHigh      402.08      (7.4%)      548.50     
(11.9%)   36.4% (  15% -   60%) 0.000
               BrowseDateTaxoFacets        5.32     (14.1%)        7.96     
(42.0%)   49.6% (  -5% -  122%) 0.000
          BrowseDayOfYearTaxoFacets        5.36     (13.8%)        8.04     
(43.7%)   50.0% (  -6% -  124%) 0.000
        BrowseRandomLabelTaxoFacets        4.30      (9.3%)        7.23     
(80.6%)   68.4% ( -19% -  174%) 0.000
              BrowseMonthTaxoFacets        5.13      (7.1%)       11.33     
(77.9%)  120.7% (  33% -  221%) 0.000
   ```
   
   # search gc stats
   | Metric | Baseline | Candidate | Difference |
   |---|---:|---:|---:|
   | Elapsed time | 17,632 ms | 15,212 ms | -2,420 ms |
   | JIT compilation time | 11,893 ms | 11,784 ms | -109 ms |
   | Young GC time | 15 ms | 12 ms | -3 ms |
   | Young GC collections | 10 | 8 | -2 |
   | Old GC time | 21 ms | 20 ms | -1 ms |
   | Old GC collections | 1 | 1 | 0 |
   | Total GC time | 36 ms | 32 ms | -4 ms |
   | Total GC share of elapsed time | 0.20% | 0.21% | +0.01 pp |
   | Average CPU load | 582.25 / 1200 | 605.78 / 1200 | +23.53 |
   
   # Indexing stats
   | Metric | Baseline | Candidate | Difference |
   |---|---:|---:|---:|
   | Total docs | 33,332,620 | 33,332,620 | 0 |
   | Indexing done time | 7,520,478 ms | 3,851,732 ms | -3,668,746 ms |
   | Wait for merges time | 708 ms | 694 ms | -14 ms |
   | Commit time | 81 ms | 17 ms | -64 ms |
   | Finished time | 7,521,314 ms | 3,852,491 ms | -3,668,823 ms |
   | Final segment count | 23 | 15 | -8 |
   | Plain text indexing throughput | 14.06 GB/hour | 27.46 GB/hour | +95.2% |
   
   Note: the final segment layout differs between the two runs, so the indexing 
throughput comparison may include merge/segment-layout effects in addition to 
the codec change.
   
   # tim store
   | Metric | Baseline | Candidate | Difference |
   |---|---:|---:|---:|
   | `.tim` files | 23 | 15 | -8 |
   | `du -sh` | 1.3G | 1.1G | ~-0.2G |


-- 
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]

Reply via email to