You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Apache Spark (JIRA)" <ji...@apache.org> on 2018/04/06 04:46:00 UTC
[jira] [Commented] (SPARK-23879) Introduce MemoryBlock API instead
of Platform API with Object
[ https://issues.apache.org/jira/browse/SPARK-23879?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16427965#comment-16427965 ]
Apache Spark commented on SPARK-23879:
--------------------------------------
User 'kiszk' has created a pull request for this issue:
https://github.com/apache/spark/pull/19222
> Introduce MemoryBlock API instead of Platform API with Object
> --------------------------------------------------------------
>
> Key: SPARK-23879
> URL: https://issues.apache.org/jira/browse/SPARK-23879
> Project: Spark
> Issue Type: Improvement
> Components: Spark Core, SQL
> Affects Versions: 2.3.0
> Reporter: Kazuaki Ishizaki
> Priority: Major
>
> This JIRA is derived from SPARK-10399.
> During the discussion, the community revealed that current Spark framework directly accesses several types of memory regions (e.g. {{byte[]}}, {{long[]}}, or {{Off-heap}}) by using {{Platform API with }}{{Object}} type. It would be good to have unified memory management API for clear memory model.
> It is also good for performance. If we can pass typed object (e.g.{{byte[]}}, {{long[]}}) to {{Platform.getXX()/putXX()}}, it is faster than using {{Platform.getXX(Object)/putXX(Object, ...)}}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org