gf2121 opened a new pull request, #13199:
URL: https://github.com/apache/lucene/pull/13199
Typically, we estimate the point value count to compare to a threshold and
all we need is just a boolean which represents whether the point count is
greater than this threshold. This pr proposes to parse the threshold into the
intersect logic and break the recursion when the threshold is exceeded.
Dynamic pruning is a case that heavily using estimate points count so i made
a try. Here is the benchmark result:
```
TaskQPS baseline StdDevQPS
my_modified_version StdDev Pct diff p-value
HighTermDayOfYearSort 374.33 (1.8%) 429.74
(0.6%) 14.8% ( 12% - 17%) 0.000
TermDTSort 388.72 (1.7%) 474.43
(1.5%) 22.1% ( 18% - 25%) 0.000
```
PS: When profiling i noticed that PointTree construction cost a lot so i
tried to make it reusable, this optimization also contributed to this speed-up.
--
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
