You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mahout.apache.org by "Hudson (JIRA)" <ji...@apache.org> on 2016/04/11 10:42:28 UTC

[jira] [Commented] (MAHOUT-1781) Dense matrix view multiplication is 4x slower than non-view one

    [ https://issues.apache.org/jira/browse/MAHOUT-1781?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15234744#comment-15234744 ] 

Hudson commented on MAHOUT-1781:
--------------------------------

FAILURE: Integrated in Mahout-Quality #3324 (See [https://builds.apache.org/job/Mahout-Quality/3324/])
MAHOUT-1781 Dense matrix view multiplication is 4x slower than non-view (smarthi: rev 5a6d140599024cd8a6bb2a4b2187376fbad40a7d)
* math/src/main/java/org/apache/mahout/math/VectorView.java
* math/src/main/java/org/apache/mahout/math/MatrixView.java
* spark/src/main/scala/org/apache/mahout/sparkbindings/SparkEngine.scala
* math-scala/src/test/scala/org/apache/mahout/math/scalabindings/RLikeVectorOpsSuite.scala
* math/src/main/java/org/apache/mahout/math/DenseVector.java
* math/src/main/java/org/apache/mahout/math/VectorBinaryAggregate.java
* math-scala/src/test/scala/org/apache/mahout/math/scalabindings/RLikeMatrixOpsSuite.scala


> Dense matrix view multiplication is 4x slower than non-view one
> ---------------------------------------------------------------
>
>                 Key: MAHOUT-1781
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-1781
>             Project: Mahout
>          Issue Type: Improvement
>    Affects Versions: 0.11.0
>            Reporter: Dmitriy Lyubimov
>            Assignee: Dmitriy Lyubimov
>            Priority: Critical
>             Fix For: 0.11.1
>
>
> if mxA and mxB are two in-core DenseMatrix matrices, then 
> mxA(::,::) %*% mxB(::,::) takes 4x the time of mxA %*% mxB.
> possibly an issue of dot products on VectorViews vs. DenseVectors.
> dot product over DenseVectors seems to not to go through aggregate() cost-optimized framework. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)