llama.cpp

b3964c1e - metal : optimize FA vec for large sequences and BS <= 8 (#15566)

metal : optimize FA vec for large sequences and BS <= 8 (#15566) * metal : optmize FA vec for large heads and sequences * metal : adjust small-batch mul mv kernels ggml-ci * batched-bench : fix total speed computation ggml-ci * cont : add comments ggml-ci