Reputation: 21
I have recently been looking into cuTENSOR (+cupy) for a speedy tensor contraction GPU library, and have been wanting to extend my single GPU code to multi GPU distributed code via cuTENSORMg;
however, I have not been able to find any pythonic support (especially through things like cupy) for cuTENSORMg, and upon doing something like
import cupyx
print(dir(cupyx))
I have only been able to find cupyx.cutensor
(single GPU presumably), and no functions to indicate multi-GPU support. I was wondering whether anyone has any tips regarding this, or whether anyone can point to some examples using cuTENSORMg in python.
Upvotes: 0
Views: 36