Anyone know how the hardware compares to the latest mac laptops for running LLMs locally? (compute and vram, shared or otherwise)