[
https://issues.apache.org/jira/browse/SINGA-88?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
wangwei resolved SINGA-88.
--------------------------
Resolution: Fixed
Assignee: wangwei
Need to optimize the memory usage (e.g., share memory space by multiple Blobs)
in the future.
Also need to optimize the training for single node and single card, in which
case we do not need to transfer gradients. The worker can update the parameters
locally in the same GPU card.
> Single node multi-cards
> -----------------------
>
> Key: SINGA-88
> URL: https://issues.apache.org/jira/browse/SINGA-88
> Project: Singa
> Issue Type: New Feature
> Reporter: Haibo Chen
> Assignee: wangwei
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
