llama.cpp

2afcdb97 - ggml-cpu: optimise s390x multiply extend instructions (#20032)