Remix.run Logo
j45 4 days ago

Not sure I can think of anything that is more performant per watt for LLMs than Apple Silicon.

saagarjha 4 days ago | parent [-]

A datacenter GPU is going to be an order of magnitude more efficient.