top | item 38419634

(no title)

thebitstick | 2 years ago

Local LLM use isn't really continuous though. You're comparing Apples to oranges.

For training especially and for the datacenter, NVIDIA cards still make absolute sense. But for consumer use cases, and even for developers, it's not so cut and dry. Very comparable performance for short-term bursts of compute.

discuss

order

No comments yet.