[
https://issues.apache.org/jira/browse/SPARK-24615?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16554327#comment-16554327
]
Thomas Graves commented on SPARK-24615:
---------------------------------------
Right so I think part of this is trying to make it more obvious to the user
what the scope actually is. This in some ways is similar to caching. I see
people many times force an evaluation after a cache to force the data to
actually be cached because otherwise it might not do what they expect. that is
why I mentioned the .eval() type functionality.
For the example:
val rddA = rdd.withResources.mapPartitions()
val rddB = rdd.withResources.mapPartitions()
val rddC = rddA.join(rddB)
Above the mapPartitions would normally get their own stages correct? So I
would think those stages would be with the resources specific but the join
would be with the default resources. then you wouldn't have to worry about
merging, etc. But you have the case with map or others where they wouldn't
normally get their own stage so the question is perhaps should they, or do you
provide something to?
> Accelerator-aware task scheduling for Spark
> -------------------------------------------
>
> Key: SPARK-24615
> URL: https://issues.apache.org/jira/browse/SPARK-24615
> Project: Spark
> Issue Type: Improvement
> Components: Spark Core
> Affects Versions: 2.4.0
> Reporter: Saisai Shao
> Assignee: Saisai Shao
> Priority: Major
> Labels: Hydrogen, SPIP
>
> In the machine learning area, accelerator card (GPU, FPGA, TPU) is
> predominant compared to CPUs. To make the current Spark architecture to work
> with accelerator cards, Spark itself should understand the existence of
> accelerators and know how to schedule task onto the executors where
> accelerators are equipped.
> Current Spark’s scheduler schedules tasks based on the locality of the data
> plus the available of CPUs. This will introduce some problems when scheduling
> tasks with accelerators required.
> # CPU cores are usually more than accelerators on one node, using CPU cores
> to schedule accelerator required tasks will introduce the mismatch.
> # In one cluster, we always assume that CPU is equipped in each node, but
> this is not true of accelerator cards.
> # The existence of heterogeneous tasks (accelerator required or not)
> requires scheduler to schedule tasks with a smart way.
> So here propose to improve the current scheduler to support heterogeneous
> tasks (accelerator requires or not). This can be part of the work of Project
> hydrogen.
> Details is attached in google doc. It doesn't cover all the implementation
> details, just highlight the parts should be changed.
>
> CC [~yanboliang] [~merlintang]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
