You are viewing a plain text version of this content. The canonical link for it is here.

Posted to commits@mxnet.apache.org by GitBox <gi...@apache.org> on 2021/09/30 08:45:10 UTC

[GitHub] [incubator-mxnet] PawelGlomski-Intel commented on a change in pull request #20589: [v1.x][BUGFIX][ONEDNN] Remove next_impl calls from deconvolution

PawelGlomski-Intel commented on a change in pull request #20589:
URL: https://github.com/apache/incubator-mxnet/pull/20589#discussion_r719191625



##########
File path: src/operator/nn/mkldnn/mkldnn_deconvolution.cc
##########
@@ -83,14 +83,12 @@ std::shared_ptr<deconv_fwd_pd_t> MKLDNNDeconvFwd::CreatePrimitiveDesc(
   const auto get_out_size     = [&pd]() { return pd->dst_desc().get_size(); };
 
   while (!ddc.CheckImplSizeReq(get_data_size(), get_weights_size(), get_out_size())) {

Review comment:
       No, ImposePlainWherePadding imposes a plain format on one descriptor at a time: first the data descriptor, then the output descriptor, and the weight descriptor at the end. This way, maybe a better implementation will be selected, than the fully plain one.




-- 
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.

To unsubscribe, e-mail: commits-unsubscribe@mxnet.apache.org

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