llama.cpp

ggml: Fix compilation issues on ARM platform when building without fp16

#10811

Merged