incubator-giraph-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Avery Ching (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (GIRAPH-28) Introduce new primitive-specific MutableVertex subclasses
Date Mon, 12 Sep 2011 02:19:08 GMT

    [ https://issues.apache.org/jira/browse/GIRAPH-28?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13102409#comment-13102409
] 

Avery Ching commented on GIRAPH-28:
-----------------------------------

Thank you for providing an interesting set of results.  I agree that this can/will be an issue
when there are lots of outgoing edges (common in real graphs).  

The last grouping in your results especially is troubling:

Tiny:   	10000	123592
Object:   	10000	4290616
Primitive:	10000	4431744

It certainly appears that the edge relationship (dest id and edge value) data structure makes
a big difference.  However, I don't understand why the object and primitive results are similar,
nor why using a TreeMap or org.apache.mahout.math.map.OpenLongFloatHashMap is so much higher
overhead than a pair of primitive arrays.  

> Introduce new primitive-specific MutableVertex subclasses
> ---------------------------------------------------------
>
>                 Key: GIRAPH-28
>                 URL: https://issues.apache.org/jira/browse/GIRAPH-28
>             Project: Giraph
>          Issue Type: New Feature
>          Components: graph
>    Affects Versions: 0.70.0
>            Reporter: Jake Mannix
>            Assignee: Jake Mannix
>         Attachments: GIRAPH-28.diff
>
>
> As discussed on the list, MutableVertex<LongWritable,DoubleWritable,FloatWritable,DoubleWritable>
(for example) could be highly optimized in its memory footprint if the vertex and edge data
were held in a form which minimized Java object usage.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Mime
View raw message