[
https://issues.apache.org/jira/browse/SPARK-18946?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15764191#comment-15764191
]
Sean Owen commented on SPARK-18946:
-----------------------------------
I'm not sure what you're proposing as a fix though -- a big object is big, yes.
It is already compressed. Does it cause a problem that is deeper than that?
> treeAggregate will be low effficiency when aggregate high dimension vectors
> in ML algorithm
> -------------------------------------------------------------------------------------------
>
> Key: SPARK-18946
> URL: https://issues.apache.org/jira/browse/SPARK-18946
> Project: Spark
> Issue Type: Improvement
> Components: ML, MLlib
> Reporter: zunwen you
> Labels: features
>
> In many machine learning algorithms, we have to treeAggregate large
> vectors/arrays due to the large number of features. Unfortunately, the
> treeAggregate operation of RDD will be low efficiency when the dimension of
> vectors/arrays is bigger than million. Because high dimension of vector/array
> always occupy more than 100MB Memory, transferring a 100MB element among
> executors is pretty low efficiency in Spark.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
