$200 'socketed' Nvidia AI GPU for servers hacked into a PCIe card with custom PCB and 3D-printed cooling — modded Tesla V100 SMX data center GPU runs AI LLMs and is more efficient than many modern midrange offerings in AI inference

A YouTuber converted a $100 Nvidia Tesla V100 server GPU with an SMX interface into a standard PCIe card for AI tasks, costing a total of $200. The GPU, featuring 16GB of HBM2 and based on the Turing architecture, was modified with a custom PCB and 3D-printed cooling for efficient performance. Testing against newer GPUs like the Radeon RX 7800 XT and RTX 3060, the V100 showcased better efficiency in AI inference tasks, despite its age. The V100 demonstrated superior performance and efficiency, making it a cost-effective option for AI applications.