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

Posted to commits@tvm.apache.org by GitBox <gi...@apache.org> on 2020/07/15 22:36:50 UTC

[GitHub] [incubator-tvm] trevor-m commented on pull request #6068: [BYOC][Optimization] Run accelerator specific optimizations

trevor-m commented on pull request #6068:
URL: https://github.com/apache/incubator-tvm/pull/6068#issuecomment-659050029


   It looks like this optimization callback is invoked after/during partitioning. What about optimizations that we want to do before annotation?
   
   For example, if the library can only support conv2d with a constant weights, we need to call bind_params_by_name and FoldConstant on the graph before annotation, so that the conv2d annotator can check if the conv2d args are constant.
   
   Another example is ConvertLayout to convert to NCHW, and then only allowing NCHW ops in the annotation.


----------------------------------------------------------------
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.

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