juliusshufan commented on issue #10921: [MXNET-500]Test cases improvement for 
MKLDNN on Gluon
URL: https://github.com/apache/incubator-mxnet/pull/10921#issuecomment-403276113
 
 
   @marcoabreu thanks for you prompt reply, I modify the cases and rerun, 
similiar issues are there.
   The GPU I used has more than 16G memory, and this might be the reason why I 
can't reproduce the issues.
   I noticed, if I did not use a smaller tensor on purpose, and:
   1. If I run the cases one by one,  like nosetests test_gluon: tc, then there 
is no issue.
   2. If I run nosetests:test_gluon, then there is no issue.
   
   The issue only happens at nosetests:test_operator_gpu, as the test_gluon.py 
is imported, the nosetests:test_operator_gpu will run many test cases. My 
assumption is, as long as the (test) purpose can be fullfiled, writhing python 
cases are fast than CPP, and since via the python API, and this is the cases I 
am writing...
   
   Limited to my knowledge on GPU, may I suggest to involve a folk have more 
GPU expertise, thanks for your help. 

----------------------------------------------------------------
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:
[email protected]


With regards,
Apache Git Services

Reply via email to