[ 
https://issues.apache.org/jira/browse/SYSTEMML-934?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Niketan Pansare updated SYSTEMML-934:
-------------------------------------
    Description: 
See org.apache.sysml.hops.ConvolutionOp's constructLops()

Though this affects Phase 1, this is not a subtask of it. It is because 
conv2d/pooling operators always fit on GPU for lenet.

  was:See org.apache.sysml.hops.ConvolutionOp's constructLops()


> Fix the forcing of convolution op to GPU
> ----------------------------------------
>
>                 Key: SYSTEMML-934
>                 URL: https://issues.apache.org/jira/browse/SYSTEMML-934
>             Project: SystemML
>          Issue Type: Bug
>            Reporter: Niketan Pansare
>
> See org.apache.sysml.hops.ConvolutionOp's constructLops()
> Though this affects Phase 1, this is not a subtask of it. It is because 
> conv2d/pooling operators always fit on GPU for lenet.



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

Reply via email to