Check which optimization flags have been used for compilation

1

Hi, is it possible to check how a pytorch package has been compiled?

I’m specifically interested in the optimization intrinsics (e.g. if SSE or AVX for x86, VSX or MMA for ppc64le etc. has been enabled)

Thanks!

2

You could get some information about lib support via print(torch.__config__.show()).

1 Like
3

Perfect - exactly what I was looking for. Thanks!