[Depurator]

I'm curious what kind of performance the Tensor G5 would have with llama.cpp, compared to a 16gb desktop gpu.