Comment by bick_nyers

You would need to fit the 389B parameters in VRAM to have a speed that is usable. Different experts are activated on a per token basis, so you would need to load/unload a large chunk of the 52B active parameters every token if you were trying to offload parameters to system RAM or SSD. PCIE 4.0 x16 speed is 64GB/s, so you can load those active parameters maybe 1 or 2 times per second, yielding an output speed of 1-2 tokens per second, which most would consider "unusable".