Ask HN: MacBook vs. Dedicated GPU for LLM
Summary
This Hacker News discussion tests the idea of using a MacBook versus a dedicated GPU for running large language models. Participants note that a MacBook with unified memory behaves like a slow GPU with ample VRAM, while dedicated GPUs have less VRAM but can run smaller models faster. Overall, both paths are described as slow with low payoff, reflecting the early stage of consumer-ready LLM inference on local hardware.