DickJC123 commented on issue #13749: Add NHWC layout support to Pooling (cpu, 
gpu cuda, gpu cuDNN)
URL: https://github.com/apache/incubator-mxnet/pull/13749#issuecomment-458645654
 
 
   Thanks for your detailed thoughts.  I view this PR as filling in a missing 
piece in the MXNet support for NHWC (with Convolution and Batchnorm already 
supporting it).  The performance benefit of having end-to-end support of NHWC 
(thereby avoiding transposes) was demonstrated for Volta hardware in 
https://mlperf.org/results/.  We're offering this benefit to upstream with this 
PR (along with others to come).  Once we complete the NHWC API support with 
this PR, I look forward to joining the discussion on how to make experimenting 
with layout easier for the user or done automatically.  

----------------------------------------------------------------
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