[GitHub] lilhope opened a new issue #8965: GPU Memory Usage keep increase (Not in train or predict)

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

lilhope opened a new issue #8965: GPU Memory Usage keep increase (Not in train or predict)
URL: https://github.com/apache/incubator-mxnet/issues/8965
 
 
   Hi:
      I extracted some features on a series of video frames. The final layer of my symbol would be something like`mx.sym.Group([x1,x2,x3,x4,x5,x6])`, then I use the `module.predict` to get the output, the output looks like `tmps = [[x1_output,x2_output,x3_output,x4_output,x5_output,x6_output],....]`, and `len(tmps)=305`, the I use the following code to get the feature:
   ```
   for tmp in tmps:
        #select the feat layer
        if some_condition:
             feat = tmp[4].asnumy()[0]
         else:
             feat = tmp[5].asnumpy()[0]
   
   ```
   but in this circulation, the GPU memory usage was keep increase, and finally raise an Error:`Out of memory`,which makes me very confuse. Any help will be appreciated.

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on 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


With regards,
Apache Git Services