llama.cpp

6232f2d7 - ggml : optimize non-SIMD Q4_0 vector dot product (#703)