[GitHub] [incubator-mxnet] leezu commented on issue #18734: Multi-gpu Mxnet training in sagemaker gives cuda error when dataloader is using multiprocessing

[GitBox <gi...@apache.org>]

leezu commented on issue #18734:
URL: https://github.com/apache/incubator-mxnet/issues/18734#issuecomment-703756493


   > We do load one tensor on GPU before we start other processes
   
   This is not safe to do in MXNet. For example, if you call `random.seed` in the child process after the parent process used cuda, it will cause segfault.
   
   You can also refer to https://github.com/apache/incubator-mxnet/issues/4659 https://github.com/apache/incubator-mxnet/issues/19291 (ie the current issue is a duplicate of the #4659)


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@mxnet.apache.org
For additional commands, e-mail: issues-help@mxnet.apache.org