llama.cpp

ba2ff79e - ggml: update comments for backends which have no memory to report (#20157)