Comment by gvand

The binary size is not really important in this case, llama.cpp should not be that far from this, what's matter as we all know is how much gpu memory we need.