Ask HN: MacBook vs. Dedicated GPU for LLM 22 days agoHow bad is the token-per-second drop-off on the Mac once you scale up the context window? Does it hit a wall at 32K or 64k tokens? 0ThreadHN