jpountz commented on PR #12489:
URL: https://github.com/apache/lucene/pull/12489#issuecomment-1717341776
I just found a bug that in practice only made BP run one iteration per
level, fixing it makes performance better (wikibigall):
```
TaskQPS baseline StdDevQPS
my_modified_version StdDev Pct diff p-value
IntNRQ 122.77 (15.4%) 114.15
(0.7%) -7.0% ( -20% - 10%) 0.363
PKLookup 294.84 (2.9%) 282.06
(2.7%) -4.3% ( -9% - 1%) 0.030
OrHighLow 713.73 (3.5%) 688.95
(3.7%) -3.5% ( -10% - 3%) 0.170
Wildcard 78.71 (4.2%) 78.01
(1.1%) -0.9% ( -6% - 4%) 0.682
Prefix3 131.65 (9.1%) 132.63
(7.3%) 0.7% ( -14% - 18%) 0.898
Respell 203.56 (0.3%) 205.74
(1.1%) 1.1% ( 0% - 2%) 0.051
HighTermMonthSort 6065.88 (2.1%) 6162.98
(1.5%) 1.6% ( -1% - 5%) 0.208
HighSpanNear 5.21 (1.7%) 5.40
(2.6%) 3.6% ( 0% - 7%) 0.021
MedSloppyPhrase 5.78 (3.5%) 6.15
(5.3%) 6.3% ( -2% - 15%) 0.047
MedSpanNear 9.40 (0.8%) 10.05
(1.1%) 6.9% ( 4% - 8%) 0.000
LowSpanNear 13.99 (1.0%) 15.28
(1.2%) 9.2% ( 6% - 11%) 0.000
HighSloppyPhrase 1.26 (4.9%) 1.38
(8.3%) 9.9% ( -3% - 24%) 0.039
OrHighHigh 46.12 (8.9%) 55.13
(6.8%) 19.5% ( 3% - 38%) 0.001
Fuzzy2 163.38 (0.8%) 199.07
(0.7%) 21.8% ( 20% - 23%) 0.000
LowSloppyPhrase 28.75 (2.2%) 35.28
(3.1%) 22.7% ( 17% - 28%) 0.000
HighPhrase 7.58 (2.1%) 9.35
(1.7%) 23.4% ( 19% - 27%) 0.000
OrHighMed 146.19 (6.5%) 183.57
(5.2%) 25.6% ( 12% - 39%) 0.000
HighTermDayOfYearSort 153.45 (2.5%) 194.38
(1.9%) 26.7% ( 21% - 31%) 0.000
Fuzzy1 259.92 (2.4%) 345.09
(2.5%) 32.8% ( 27% - 38%) 0.000
HighTerm 478.18 (9.8%) 670.01
(9.2%) 40.1% ( 19% - 65%) 0.000
MedTerm 577.98 (9.0%) 845.32
(10.0%) 46.3% ( 25% - 71%) 0.000
AndHighMed 157.39 (4.5%) 243.75
(7.3%) 54.9% ( 41% - 69%) 0.000
LowTerm 1016.15 (7.6%) 1671.11
(9.8%) 64.5% ( 43% - 88%) 0.000
AndHighLow 746.14 (1.7%) 1227.66
(4.2%) 64.5% ( 57% - 71%) 0.000
MedPhrase 41.72 (2.0%) 71.95
(3.4%) 72.4% ( 65% - 79%) 0.000
AndHighHigh 31.03 (7.0%) 56.59
(13.4%) 82.4% ( 57% - 110%) 0.000
LowPhrase 69.04 (1.5%) 126.15
(3.4%) 82.7% ( 76% - 88%) 0.000
```
Space savings are also bigger on postings:
| File | before (MB) | after (MB) |
| - | - | - |
| terms (tim) | 767 |763 |
| postings (doc) | 2779 | 2260 |
| positions (pos) | 11356 | 10522 |
| points (kdd) | 100 | 99 |
| doc values (dvd) | 456 | 462 |
| stored fields (fdt) | 249 | 226 |
| norms (nvd) | 13 | 13 |
| total | 15734 |14360 |
--
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]
