[GitHub] [incubator-mxnet] shesung commented on issue #11076: [MXNET-491] Use depthwise convolution by cuDNNv7 if available, updated version

Tue, 04 Sep 2018 18:19:11 -0700 

I observed barely no improvement when using mxnet 1.2.1 + cuda8 + cudnn 7.2.1 
on 1080ti 
When setting MXNET_CUDNN_AUTOTUNE_DEFAULT=0, performance drop rapidly.

[ Full content available at: 
https://github.com/apache/incubator-mxnet/pull/11076 ]
This message was relayed via gitbox.apache.org for [email protected]

Reply via email to