llama.cpp

cuda : add missing support check for xielu

#17895

Merged