llama.cpp

ggml : move 32-bit arm compat in ggml-impl.h

#6865

Merged