Prerequisites

• I am running the latest code. Mention the version if possible as well.
• I carefully followed the README.md.
• I searched using keywords relevant to my issue to make sure that I am creating a new issue that is not already open (or closed).
• I reviewed the Discussions, and have a new and useful enhancement to share.

Feature Description

llama.cpp already has the --prio N and --prio-batch N options to allow specifying a higher-than-normal priority. I'd suggest to extend them to allow using an idle (lowest) priority as well. For example, the help descriptions for those options could be changed to: set process/thread priority : idle(-2), low(-1), normal(0), medium(1), high(2), realtime(3).

• -2 could mean a niceness value of 19

As I don't see the harm in allowing an idle (lowest) priority, I haven't posted this in the discussions first. Hopefully that's understandable.

Motivation

This could help with running AI models on workstations where you only use the CPU for inferencing (for example because the system only has an older built-in GPU).

The use case: so your computer stays more responsive during inferencing, so the browser and file manager and so on are less laggy during inferencing.

Allowing llama.cpp to run with a niceness of 19 via the existing command-line options would be helpful.

Possible Implementation

No response