> hands down the fastest GPU ever for local LLM purposes Not really. On pure speed a 5090 will still comfortably beat a top end Mac configuration. Just constra

> hands down the fastest GPU ever for local LLM purposes Not really. On pure speed a 5090 will still comfortably beat a top end Mac configuration. Just constrained to 32gb mem ofc Top Mac has mem throughput around the 900 mark while a 5090 is at 1800