llama.cpp

f5fddb6d - ggml-cuda : remove some superfluous comments for TQ2_0 tile loading