Quote:
Originally Posted by BradZax
[You must be logged in to view images. Log in or Register.]
So what's the deal with this new Nvidia computer? My understanding is that it.... does a local LLM?
It uses its ow2n processor and can run offline? I am having a hard time understanding what they're selling.
|
yeah it(DGX Spark) runs big local models well due to having a shit load of vram, but less CUDA cores than a high end gaming GPU and the high end gaming GPU doesn't have enough ram on the card to run big models, so idk where the sweet spot is going to land on all this shit yet for like my use case with Kaia
i'm still looking into used server-grade gpu cards with a pci adapter and 3d printed cooling shroud to work in a desktop