llama.cpp

4df6e859 - cuda : add missing support check for xielu (#17895)