when I run predict with dnn in cpu pattern, I got a very high utilization, It seems occupy several cpu core,I don‘t know why，If you give any helps I will thanks very much!
Hi,
CPU implementation uses OpenMP to paralellize computations over multiple cores to speed up computations.
You can use torch.get_num_threads() and torch.set_num_threads() to see how many threads (and thus core) it will use and to set a limit if you want it to use less.