dual GPU split inference with PyTorch

[localcraft17]

Posting this in case it saves someone else a little time.

If you were starting from scratch today for this use case, what would you choose first and what would you ignore?

I'm especially interested in mistakes worth avoiding.

[embersmith]

What exact setup are you running? Is this local-only or are you routing anything externally?

[turbogrid11]

Interesting, I hadn't tried that. The pattern around tokens/sec has been pretty similar for me.