llama.cpp

c8edceb0 - ggml-cuda: hip add specific config table for AMD GCN (#27841)