-
|
And if it is possible would it lead to a speedup? |
Beta Was this translation helpful? Give feedback.
Answered by
ericspod
Sep 16, 2023
Replies: 1 comment
-
|
|
Beta Was this translation helpful? Give feedback.
0 replies
Answer selected by
grudloff
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Spacingwill work using CUDA if the input tensor is on a GPU device already. It should give a speed up if the rest of your transform sequence is optimal, the usual places where time is list is converting tensors or moving them to and from devices so avoiding these is important.