V100

llama.cpp Multi-GPU Guide: Offload, Split Mode, tensor-split, and VRAM