GGUF vs bin?

[VadimBoev]

Hi everybody. I'm sorry if I asked an incorrect question, I'm not good at this.
I am interested in maximum performance. Which is better than GGUF or bin?

Also, I saw a topic in which ggerganov advised to disable map and compile llama.cpp after such changes.
Does it really work? And how can I set parameters for high performance?
All I managed to do was launch "./main -m ../openchat-3.5-0106.Q4_K_M.of "Building a website can be done in 10 simple steps:\step 1:" -n 1024 -e"

[Jeximo]

I am interested in maximum performance. Which is better than GGUF or bin

Same performance.

Also, I saw a topic in which ggerganov advised to disable map and compile llama.cpp after such changes.
Does it really work?

--no-mmap in the command line may help performance, depending on your system RAM --no-mmap

And how can I set parameters for high performance?

Add --threads N, -t N, --threads N

If you build including GPU, then also add -ngl N -ngl N, --n-gpu-layers N

[VadimBoev]

thank you!