[ 
https://issues.apache.org/jira/browse/MAHOUT-1921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15852972#comment-15852972
 ] 

ASF GitHub Bot commented on MAHOUT-1921:
----------------------------------------

Github user andrewpalumbo commented on a diff in the pull request:

    https://github.com/apache/mahout/pull/274#discussion_r99477184
  
    --- Diff: 
math-scala/src/main/scala/org/apache/mahout/math/decompositions/DSSVD.scala ---
    @@ -22,13 +22,18 @@ object DSSVD {
        * @return (U,V,s). Note that U, V are non-checkpointed matrices (i.e. 
one needs to actually use them
        *         e.g. save them to hdfs in order to trigger their computation.
        */
    -  def dssvd[K](drmA: DrmLike[K], k: Int, p: Int = 15, q: Int = 0):
    +  def dssvd[K](drmA: DrmLike[K],
    +               k: Int,
    +               p: Int = 15,
    +               q: Int = 0,
    +               cacheHint: CacheHint.CacheHint = CacheHint.MEMORY_ONLY):
    --- End diff --
    
    Not sure what we should make the default `CacheHint` here and other places, 
e.g. #276.
    
    for `drmDfsRead(...)` we use `CacheHint.NONE` but that may be something 
that we should let the user supply as well.  Most parallaization methods use 
`CacheHint.NONE`.  I would think that if we have anything hardcoded or as a 
default, we should use `CacheHint.MEMORY_AND_DISK` since we have no idea of the 
size of the user's data..  but I suppose that is making Assumptions as well.


> DSSVD Propagates cache hint
> ---------------------------
>
>                 Key: MAHOUT-1921
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-1921
>             Project: Mahout
>          Issue Type: Improvement
>          Components: Math
>    Affects Versions: 0.13.0
>            Reporter: Trevor Grant
>            Assignee: Trevor Grant
>            Priority: Minor
>              Labels: beginner
>             Fix For: 0.13.0
>
>
> The DSSVD does lots of check pointing, but currently only the default 
> checkpoint cacheHint is given.  
> The user should be able to pass a checkpointing hint as this can lead to 
> dramatic performance issues in some cases.
> https://github.com/apache/mahout/blob/master/math-scala/src/main/scala/org/apache/mahout/math/decompositions/DSSVD.scala



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Reply via email to