[
https://issues.apache.org/jira/browse/LUCENE-7259?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Robert Muir updated LUCENE-7259:
--------------------------------
Attachment: LUCENE-7259.patch
Here is a patch. Our benchmark is fairly noisy here, but it seems consistently
slightly faster for BoxQuery over multiple runs.
{noformat}
before:
BEST M hits/sec: 83.63882053405949
BEST QPS: 85.10687863474622
BEST M hits/sec: 85.81999125313045
BEST QPS: 87.32633403218385
BEST M hits/sec: 84.68474195132809
BEST QPS: 86.17115843391811
after:
BEST M hits/sec: 87.43648355453598
BEST QPS: 88.97119957705006
BEST M hits/sec: 87.06546761099004
BEST QPS: 88.59367143070249
BEST M hits/sec: 87.46697895621536
BEST QPS: 89.00223024478392
{noformat}
Even if its not a huge speedup, its clearly removing stuff from the super hot
loop, which is what we should do!
> speed up MatchingPoints cost() estimation
> -----------------------------------------
>
> Key: LUCENE-7259
> URL: https://issues.apache.org/jira/browse/LUCENE-7259
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Robert Muir
> Attachments: LUCENE-7259.patch
>
>
> MatchingPoints currently tracks a counter in the super-hot add() loop. While
> not a big deal, we can easily just use the grow() api for this instead (which
> is only currently called e.g. every 1k docs).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
