Ask HN: MacBook vs. Dedicated GPU for Local LLM Inference

Hacker News·low signal

An Ask HN thread (29 points, 54 comments) weighs running local LLMs on Apple Silicon unified memory versus a dedicated discrete GPU, surfacing practitioner trade-offs on memory bandwidth, VRAM ceilings, model sizes, and cost. The discussion is a useful pulse-check on where hobbyist and indie-builder local-inference sentiment sits as open-weight models close the gap with closed frontier ones. Community-driven, hands-on perspectives rather than vendor benchmarks.