Arch Linux system optimization for Audio Work (SuperCollider/Pure Data) ?

[uptick]

Just wondering if anyone have experience making music with Arch Linux and how to best setup/optimize the system for music production?

I've only used Arch Linux for regular every day stuff in the past, so I'm a complete novice when it comes to the audio department. But, lately, I've been really interested in learning SuperCollider and Pure Data, so I figured I would ask here for some advice on how to properly set up everything.

I did read through https://wiki.archlinux.org/index.php/Professional_audio but are all these steps necessary and vital to increasing performance and decreasing latency? How much difference do they make?

Anyways, as a newbie, I would love to hear about your own experience and what you would consider to be the most important settings when it comes to fine tuning your Arch Linux for music production.

[Jack Winter]

IMO, install a rt kernel and the realtime-privileges package, and use rtprio to raise the priority of your soundcard. Take much of the other rest of the advice with a pinch of salt.. Yes it might be useful to change the cpu powersave governor too. Start with that and then ask for more advice if you have any specific problems.

[uptick]

Thank you so much Jack,
By the way, what's a good latency to shoot for?
What range would you consider bad, average, good, excellent?

Also, I noticed that when I use jack via (Qjackctl), I'm unable to have sound from other applications such as browser, mpv, etc.
Is there a way to still ahve sound working for everything else while working with jack?

Currently I only have alsa. I don't even have pulseaudio installed as I feared it might interfere with jack somehow and complicate things.
Do I need to install pulseaudio and pulseaudio-alsa?

[CrocoDuck]

Hi there, I am on Arch too.

I did read through https://wiki.archlinux.org/index.php/Professional_audio but are all these steps necessary and vital to increasing performance and decreasing latency? How much difference do they make?

It really depends on your system. In my experience setting noatime in fstab really helps, as well as rtirq, especially for USB devices. Typically, what I found most useful is what comes from realtimeconfigquickscan tips. You can find it on the AUR.

I did a tutorial for ArchBang a lot of time ago. Beware, it is outdated. However, the amount of actions you might need to do is perhaps similar, but don't be scared. Once you configured your system for audio a couple of times it you get used to it and it really takes 20 minutes to do. It is not as hard as it looks.

My tip is: use realtimeconfigquickscan to make sure all the mundane things are sorted (you are in the audio group, noatime is set, rtirq is in...) then use the heavy artillery (rt kernels, killing processes and services...) if you need to.

By the way, what's a good latency to shoot for?
What range would you consider bad, average, good, excellent?

I did a series of articles on my blog. You can find a table of perceptual thresholds towards the bottom of this long post. Turns out that latency perception depends on what kind of instruments you are used to play, for various physical and psychophysiological reasons. All information comes from a scientific paper.

If you play guitar I would say, based on that information:

Excellent: <= 5 ms. (this should only change the "quality" of the sound)
Good: ~10 ms.
Average: ~15 ms (from here you start hearing a delay)
Bad: >20 ms (sounds more and more as a delay)

I have been playing for years with some 13 ms of round-trip latency, and I found that to work OK. By the way, here how to measure latency on Linux.