llama.cpp

f12ceaca - ggml-ci : try to improve build time (#9160)