I have an old multi-GPU node lying around at work...
A user on Reddit is seeking advice on how to repurpose an underutilized multi-GPU server at their workplace for local AI inference. The server is equipped with 8 NVIDIA Quadro RTX 6000 GPUs, 192 GB of VRAM, 512 GB of RAM, and 112 CPU threads. The user wants to make a business case to their employer for this repurposing and is asking for recommendations on which types of AI models could be run on this hardware that would offer significant advantages over single-GPU setups. AI
