lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Robert Muir (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (LUCENE-7215) don't invoke full haversin for LatLonPoint.newDistanceQuery
Date Wed, 13 Apr 2016 22:16:25 GMT

     [ https://issues.apache.org/jira/browse/LUCENE-7215?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Robert Muir updated LUCENE-7215:
--------------------------------
    Attachment: LUCENE-7215.patch

Here is a patch. I see these numbers (i ran 10 iterations x 3 jvms for each and numbers are
consistent):

2D distance with no filter: 33.7 QPS -> 35.4 QPS
2D distance with 0.0001% filter: 53.0 QPS -> 58.0 QPS


> don't invoke full haversin for LatLonPoint.newDistanceQuery
> -----------------------------------------------------------
>
>                 Key: LUCENE-7215
>                 URL: https://issues.apache.org/jira/browse/LUCENE-7215
>             Project: Lucene - Core
>          Issue Type: Improvement
>            Reporter: Robert Muir
>         Attachments: LUCENE-7215.patch
>
>
> For tree traversals and edge cases we still sometimes invoke full haversin (with asin()
call and everything). this is not necessary: we just need to compute the exact sort key needed
for comparisons.
> While not a huge optimization, its obviously less work and keeps the overhead of the
BKD traversal as low as possible. And it removes the slow asin call from any hot path (its
already done for sorting too), with its large tables and so on.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org


Mime
View raw message