llama.cpp

df17a4c9 - CUDA: fix FA FP16 accumulator overflow for Granite (#18614)