Hacker News new | ask | show | jobs
by om8 314 days ago
Sure, but integrated graphics usually lacks vram for LLM inference.
1 comments

Which means that inference would be approximately the same speed (but compute offloaded) as the suggested CPU inference engine.