yangulei commented on PR #11513: URL: https://github.com/apache/tvm/pull/11513#issuecomment-1150882679
@billishyahao I remember that you introduced a mechanism to query optimal layout and alter op layout before the graph is consumed by DNNL, just like we do for `Conv`. Did you cancel that and do the layout transform in dnnl json runtime? If so, I agree with @apeskov and wondering why no layout transform of the weights. -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
