llama.cpp
ggml-cpu: Faster IQ1 mul_mat_vec on AVX2 using BMI2 instructions
#12154
Merged
Merged