llama.cpp
ggml : use dynamic thread scheduling for matrix multiplication
#6915
Merged
Merged