Wnat pytorch compile with ICC

I have GPU server and via nvidia-docker installed the pytorch run the pytorch test related scripts also. But want to measure the performance of GPU server with intel parallel studio (ICC).

Please guide how to use ICC(intel c++ compiler) with pytorch to check performance on gpu server.

you’re on your own here. we dont use any non-standard build tools so you should hopefully figure it out.

can you tell me which standard build tools you used for the permormance improvement ???