Selecting the best GPU server for ChatGPT inference depends on balancing VRAM capacity, tensor core performance, me