Tvm Inference peformance almost 10x better than pytorch

Questions
2

module.run in non-blocking. it need to put timer.end after module.get_output

1 Like