systemml-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Matthias Boehm (JIRA)" <j...@apache.org>
Subject [jira] [Closed] (SYSTEMML-1967) Spark rand/seq instructions generate too many partitions
Date Thu, 30 Nov 2017 22:16:01 GMT

     [ https://issues.apache.org/jira/browse/SYSTEMML-1967?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Matthias Boehm closed SYSTEMML-1967.
------------------------------------
       Resolution: Fixed
         Assignee: Matthias Boehm
    Fix Version/s: SystemML 1.0

> Spark rand/seq instructions generate too many partitions 
> ---------------------------------------------------------
>
>                 Key: SYSTEMML-1967
>                 URL: https://issues.apache.org/jira/browse/SYSTEMML-1967
>             Project: SystemML
>          Issue Type: Bug
>            Reporter: Matthias Boehm
>            Assignee: Matthias Boehm
>             Fix For: SystemML 1.0
>
>
> The existing spark rand instructions makes use of an optimizer utils methods for estimating
the total size under awareness of sparsity in order to determine the number of partitions.
Since this methods was later overloaded, this instructions currently calls the wrong primitive
for sparsity with the nnz leading to dense estimates.



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

Mime
View raw message