lichun-wang commented on issue #5629: URL: https://github.com/apache/incubator-tvm/issues/5629#issuecomment-631917513
> Use debugger runtime to see which layer is the bottleneck? i print the time of the resnest, I found the cost of first layer( just conv + bn + relu ) was bigger than resnet, resnet just cost 0.75ms, resnest cost 37ms, the only difference of is channel resnest is 64,resnet is 32 . I open a issus on disscuss.tvm.ai :https://discuss.tvm.ai/t/question-after-auto-tune-the-speed-of-resnest101-and-resnet101-have-big-difference/6748 ---------------------------------------------------------------- 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: [email protected]
