llama.cpp

0bf47a1d - server: add memory breakdown print (#16740)