llama.cpp

compare-llama-bench.py: fix long hexsha args

#6424

Merged