llama-bench: introduce -hf and -hff flags & use --mmap 1 by default
#20211
Loading
-hf and -hff flags & use --mmap 1 by default
#20211