llama.cpp

ggml-cpu: Don't pass `-mpowerpc64` when `-mcpu` already implies it

#17308

Merged