User asks about repurposing old multi-GPU node for local LLM inference

A Reddit user seeks advice on repurposing an 8x Quadro RTX 6000 GPU node (192GB VRAM) for local LLM inference. The post asks for model recommendations and software setup tips.