llama.cpp

ggml : add AVX support based on AVX2 code

#1430

Merged