[ 
https://issues.apache.org/jira/browse/MAHOUT-1190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13628298#comment-13628298
 ] 

Robin Anil edited comment on MAHOUT-1190 at 4/10/13 9:42 PM:
-------------------------------------------------------------

Please add a test case to VectorBenchmarks.java, Most of the code in Mahout are 
tuned with RandomAccess and DenseVectors in mind. Sequential access is a slow 
format. And really the lower memory requirements does not really justify its 
existence. I've always wanted to delete that one from the codebase altogether. 
Vector benchmarks tests a lot of combination of things. I wouldnt want other 
formats to regress on performance.
                
      was (Author: robinanil):
    Please add a test case to VectorBenchmarks.java, Most of the code in Mahout 
is runed for RandomAccess and DenseVectors in mind. Sequential access is a slow 
format. And really the lower memory requirements does not really justify its 
existence. I would have wanted to delete that one from the codebase altogether. 
Vector benchmarks tests a lot of combination of things. I wouldnt want other 
formats to regress on performance.
                  
> SequentialAccessSparseVector function assignment is very slow
> -------------------------------------------------------------
>
>                 Key: MAHOUT-1190
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-1190
>             Project: Mahout
>          Issue Type: Bug
>            Reporter: Dan Filimon
>
> Currently when calling .assign() on a SASV with another vector and a custom 
> function, it will iterate through it and assign every single entry while also 
> referring it by index.
> This makes the process *hugely* expensive. (on a run of BallKMeans on the 20 
> newsgroups data set, profiling reveals that 92% of the runtime was spent 
> updating assigning the vectors).
> Here's a prototype patch:
> https://github.com/dfilimon/mahout/commit/63998d82bb750150a6ae09052dadf6c326c62d3d

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to