llama.cpp

llama : report compute graph reallocations during inference

#30329

Merged