[jira] [Updated] (TEZ-3777) Avoid buffer copies by passing RLE flag to TezMerger from PipelinedSorter

2017-06-27 Thread Rajesh Balamohan (JIRA)

 [ 
https://issues.apache.org/jira/browse/TEZ-3777?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Rajesh Balamohan updated TEZ-3777:
--
Attachment: TEZ-3777.2.patch

> Avoid buffer copies by passing RLE flag to TezMerger from PipelinedSorter
> -
>
> Key: TEZ-3777
> URL: https://issues.apache.org/jira/browse/TEZ-3777
> Project: Apache Tez
>  Issue Type: Bug
>Reporter: Rajesh Balamohan
>Assignee: Rajesh Balamohan
> Attachments: TEZ-3777.1.patch, TEZ-3777.2.patch
>
>
> RLE information computed in pipelinedSorter can be passed on to 
> {{TezMerger.merge}} in  {{PipelinedSorter::flush}}. Depending on dataset, 
> this can save lot of data copies.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)


[jira] [Updated] (TEZ-3777) Avoid buffer copies by passing RLE flag to TezMerger from PipelinedSorter

2017-06-27 Thread Rajesh Balamohan (JIRA)

 [ 
https://issues.apache.org/jira/browse/TEZ-3777?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Rajesh Balamohan updated TEZ-3777:
--
Attachment: TEZ-3777.1.patch

> Avoid buffer copies by passing RLE flag to TezMerger from PipelinedSorter
> -
>
> Key: TEZ-3777
> URL: https://issues.apache.org/jira/browse/TEZ-3777
> Project: Apache Tez
>  Issue Type: Bug
>Reporter: Rajesh Balamohan
>Assignee: Rajesh Balamohan
> Attachments: TEZ-3777.1.patch
>
>
> RLE information computed in pipelinedSorter can be passed on to 
> {{TezMerger.merge}} in  {{PipelinedSorter::flush}}. Depending on dataset, 
> this can save lot of data copies.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)