llama.cpp

cmake: Added more x86_64 CPU backends when building with `GGML_CPU_ALL_VARIANTS=On`

#18186

Merged