cross-posted from: lemmy.world/post/51495849
Latest rumor is that NVIDIA is making too much with AI cards. They won’t release a new consumer line for a couple more years.
It isn’t worth spending any waffers on the high end nodes for chips that offer such relatively low margins.
SnotFlickerman@lemmy.blahaj.zone 2 weeks ago
*laughs in AMD
(I know they’re not that much better but they’re so much more well supported in Linux it’s absurd. NVIDIA only started getting more support that was focused on AI, which is cool and all, local AI is all right, but no very useful in everyday context.)
daannii@lemmy.world 2 weeks ago
I switched from Nvidia and Intel to Amd Amd for my newer PC. Couldn’t be happier. No issues. Better prices too.
AMD has software for both its processors and gpus. So nothing to trade off as far as I can tell.
DrunkenDuckling@lemmus.org 2 weeks ago
Fr, how anyone still pays extra for Nvidia’s bullshit is crazy to me.
surewhynotlem@lemmy.world 2 weeks ago
How’s the local LLM support on Linux with AMD?
(Yes I could google, but I think people here would give more realistic answers at a more reasonable tech skill level)
ozymandias117@lemmy.world 2 weeks ago
llama.cpp (and my extension Ollama) have a Vulkan backend now, which works fine for my usecases on AMD