Possible to use an Nvidia GPU with Framework 16?

I’d like to work with the Cuda Nvidia framework to do ML/AI stuff, and that framework requires Nvidia GPU’s.

1 Like

I’ve got an nvidia gpu plugged in via thunderbolt:

How about not using cuda. people defaulting to cuda is why we don’t have good enough competition. and why nvidia tries to have us over a barrel. if developers would revolt against cuda and use more hardware agnostic programing we can get better results.


Especially with zluda being a thing that Nvidia is taking steps to squish.

Idealistically, I agree. We both know it’s not that simple in practice though.