[
https://issues.apache.org/jira/browse/LUCENE-1604?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12698880#action_12698880
]
Shon Vella commented on LUCENE-1604:
------------------------------------
The check isn't for free because the HotSpot compiler doesn't check every
reference for null, it just traps the resulting segmentation fault that
dereferencing a null pointer causes. We could leverage this by not checking for
null, but catch a null pointer exception, though I believe the extra overhead
of a try/catch block would be more expensive than just checking.
The performance hit is likely only in the case where it isn't equal null - in
the equal null case you end up saving a multiply plus a memory reference that
is likely going to cause many CPU cache faults over the coarse of a search long
enough to matter, which together probably add up to more than the cost of the
check.
An alternate approach that would eliminate this overhead is to subclass the
scorers that use norms and create an appropriate scorer that doesn't require
the check. The drawback of this approach would be that it is harder to maintain.
> Stop creating huge arrays to represent the absense of field norms
> -----------------------------------------------------------------
>
> Key: LUCENE-1604
> URL: https://issues.apache.org/jira/browse/LUCENE-1604
> Project: Lucene - Java
> Issue Type: Improvement
> Components: Index
> Affects Versions: 2.9
> Reporter: Shon Vella
> Priority: Minor
> Fix For: 2.9
>
> Attachments: LUCENE-1604.patch
>
>
> Creating and keeping around huge arrays that hold a constant value is very
> inefficient both from a heap usage standpoint and from a localility of
> reference standpoint. It would be much more efficient to use null to
> represent a missing norms table.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: java-dev-h...@lucene.apache.org
