giraph-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Claudio Martella (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (GIRAPH-459) Group Vertex Mutations by Partition ID
Date Fri, 04 Jan 2013 15:44:12 GMT

    [ https://issues.apache.org/jira/browse/GIRAPH-459?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13543959#comment-13543959
] 

Claudio Martella commented on GIRAPH-459:
-----------------------------------------

Running PageRankBenchmark with 3 workers on my laptop, 100k vertices, 100 edges each, HashMapVertex
and PseudoRandomEdgeInputFormat.

trunk:
Superstep 3 (milliseconds)	6,025	0	6,025
Superstep 4 (milliseconds)	5,605	0	5,605
Superstep 20 (milliseconds)	124	0	124
Setup (milliseconds)	6,025	0	6,025
Shutdown (milliseconds)	154	0	154
Superstep 7 (milliseconds)	5,691	0	5,691
Superstep 16 (milliseconds)	5,605	0	5,605
Superstep 9 (milliseconds)	5,571	0	5,571
Superstep 8 (milliseconds)	10,527	0	10,527
Superstep 14 (milliseconds)	10,538	0	10,538
Input superstep (milliseconds)	15,764	0	15,764
Superstep 6 (milliseconds)	11,103	0	11,103
Superstep 18 (milliseconds)	5,623	0	5,623
Superstep 5 (milliseconds)	10,666	0	10,666
Superstep 1 (milliseconds)	14,376	0	14,376
Total (milliseconds)	186,270	0	186,270
Superstep 15 (milliseconds)	5,609	0	5,609
Superstep 17 (milliseconds)	5,825	0	5,825
Superstep 12 (milliseconds)	5,796	0	5,796
Superstep 10 (milliseconds)	6,037	0	6,037
Superstep 19 (milliseconds)	10,729	0	10,729
Superstep 11 (milliseconds)	10,875	0	10,875
Superstep 0 (milliseconds)	11,098	0	11,098
Superstep 2 (milliseconds)	11,221	0	11,221
Superstep 13 (milliseconds)	5,673	0	5,673
GIRAPH-459:
Superstep 3 (milliseconds)	5,845	0	5,845
Superstep 4 (milliseconds)	5,912	0	5,912
Superstep 20 (milliseconds)	145	0	145
Setup (milliseconds)	6,165	0	6,165
Shutdown (milliseconds)	65	0	65
Superstep 7 (milliseconds)	10,625	0	10,625
Superstep 16 (milliseconds)	6,031	0	6,031
Superstep 9 (milliseconds)	5,951	0	5,951
Superstep 8 (milliseconds)	5,626	0	5,626
Superstep 14 (milliseconds)	10,653	0	10,653
Input superstep (milliseconds)	11,213	0	11,213
Superstep 6 (milliseconds)	5,876	0	5,876
Superstep 18 (milliseconds)	10,719	0	10,719
Superstep 5 (milliseconds)	11,108	0	11,108
Superstep 1 (milliseconds)	14,978	0	14,978
Total (milliseconds)	186,255	0	186,255
Superstep 15 (milliseconds)	10,663	0	10,663
Superstep 17 (milliseconds)	10,914	0	10,914
Superstep 12 (milliseconds)	5,636	0	5,636
Superstep 10 (milliseconds)	6,013	0	6,013
Superstep 19 (milliseconds)	11,421	0	11,421
Superstep 11 (milliseconds)	5,559	0	5,559
Superstep 0 (milliseconds)	12,874	0	12,874
Superstep 2 (milliseconds)	6,656	0	6,656
Superstep 13 (milliseconds)	5,602	0	5,602


Looks good. I fixed the rest of the issues. Waiting for a comment on (1).
                
> Group Vertex Mutations by Partition ID
> --------------------------------------
>
>                 Key: GIRAPH-459
>                 URL: https://issues.apache.org/jira/browse/GIRAPH-459
>             Project: Giraph
>          Issue Type: Improvement
>          Components: graph
>            Reporter: Claudio Martella
>         Attachments: GIRAPH-459.patch
>
>
> Currently, vertex mutations, and implicit creations of vertices based on messages to
non-existing vertices, are executed randomly partition-wise. The iterated vertices can belong
to different partitions. This is bad when we work out-of-core, as we need to load and unload
the whole partition for each vertex. We should group these operations per-partition, and batch-execute
them.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message