llama.cpp

ggml: aarch64: implement mmla kernels for q8_0_q8_0, q4_0_q8_0 and q4_1_q8_1 quantized gemm

#4966

Merged