llama.cpp

0e0590ad - cuda : update supports_op for matrix multiplication (#8245)