You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@singa.apache.org by "wangwei (JIRA)" <ji...@apache.org> on 2017/04/07 07:07:41 UTC

[jira] [Commented] (SINGA-308) CPU-GPU parallelism

    [ https://issues.apache.org/jira/browse/SINGA-308?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15960387#comment-15960387 ] 

wangwei commented on SINGA-308:
-------------------------------

Yes. It is possible.
To use a GPU and a CPU to do the training, just remove device2 related lines, including 
https://github.com/apache/incubator-singa/blob/master/examples/cifar10/alexnet-parallel.cc#L224
https://github.com/apache/incubator-singa/blob/master/examples/cifar10/alexnet-parallel.cc#L227
https://github.com/apache/incubator-singa/blob/master/examples/cifar10/alexnet-parallel.cc#L230
https://github.com/apache/incubator-singa/blob/master/examples/cifar10/alexnet-parallel.cc#L236
https://github.com/apache/incubator-singa/blob/master/examples/cifar10/alexnet-parallel.cc#L237

> CPU-GPU parallelism 
> --------------------
>
>                 Key: SINGA-308
>                 URL: https://issues.apache.org/jira/browse/SINGA-308
>             Project: Singa
>          Issue Type: Test
>          Components: Core, PySINGA
>         Environment: Ubuntu 16.04
> CPU-GPU of the same machine
>            Reporter: Muhammad Hamdan
>              Labels: test
>
> Is it possible to parallelize the alexnet model for the cifar10 example on a CPU and GPU instead of 2-GPUs ? Assuming asynchronous communication between the two components



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