llama.cpp

Fix loongarch code dependency on O2 compilation options in ggml-quants

#7636

Merged