• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

B580 vs RX 7600 vs RTX 4060 in Pytorch/Tensorflow (AI) benchmarks?

it sounds like there are major hurdles from using non-nvidia hardware with AI libraries.
That's correct. Like @bug suggested, implement some workloads on various APIs
 
Intel and amd consumer cards are really focused on being gaming cards. Though I hear AMD is planning to merge the two divisions soon so they release something similar to nvidia. Same silicon for consumer and enterprise products, with the draw for enterprise most likely being more vram, or maybe hbm or whatever other crazy stuff goes on in highest price enterprise tiers.

Anyway, yeah for now I mean, the 4060 is pretty weak but I still think its probably the best option if thats all you can afford, if nothing else, then for ease of use. I'm not an expert on the matter though. Just what I've heard.

Its not that you can't get other cards to do the work, its just a lot more work to get them to do that work, and I believe there is an overhead for that. Nvidia has something of monopoly on the software side of ai....unfortunately.

-------
You can tell clear as day now by the 32gb of vram on nvidias 5090s that they now want even want their gaming line to also do ai, as that much vram is useless right now and will be for some time for gaming, even 24gb is still overkill. For ai though, more vram is always welcome.
 
Back
Top