Only slightly related question: is there such a thing as an external nVidia GPU for AI models? I know I can rent cloud GPUs but I am wondering if long-term something like an external GPU might be worth it.
Generally speaking, buying outright is always cheaper than renting, because you can always continue to run the device potentially for years, or sell it to reclaim some capital.
A 3090 (used) is the best bang for your buck for any LLM / StableDiffusion work right now. I’ve seen external GPU enclosures, though they probably cost as much as slapping a used 3090 into a barebones rig and running it headless in a closet.
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: [email protected]
No game suggestions, friend requests, surveys, or begging.
No Let’s Plays, streams, highlight reels/montages, random videos or shorts.
No off-topic posts/comments, within reason.
Use the original source, no clickbait titles, no duplicates.
(Submissions should be from the original source if possible, unless from paywalled or non-english sources.
If the title is clickbait or lacks context you may lightly edit the title.)
Only slightly related question: is there such a thing as an external nVidia GPU for AI models? I know I can rent cloud GPUs but I am wondering if long-term something like an external GPU might be worth it.
Generally speaking, buying outright is always cheaper than renting, because you can always continue to run the device potentially for years, or sell it to reclaim some capital.
A 3090 (used) is the best bang for your buck for any LLM / StableDiffusion work right now. I’ve seen external GPU enclosures, though they probably cost as much as slapping a used 3090 into a barebones rig and running it headless in a closet.