llama.cpp

ggml webgpu: minor set rows optimization

#16810

Merged