| Author |
Message |
finotti
|
|
Post subject: [SOLVED] low latency kernel
 Posted: 26.01.2017, 06:18
|
|
Joined: 2010-09-12
Posts: 493
Status: Offline
|
|
I'm trying to do some recordings using Ardour, and from what I've been reading, one needs a "low latency kernel". So, I thought I would ask if aptosid's kernel is low latency (or if it is really necessary a special kernel).
(Sorry for the multiple recent posts, but this process of getting to record hasn't been easy and is totally new to me...) |
Last edited by finotti on 04.02.2017, 14:55; edited 1 time in total
|
| |
|
|
|
 |
|
slh
|
|
|
Post subject: RE: low latency kernel
Posted: 26.01.2017, 11:50
|
|
Joined: 2010-08-25
Posts: 962
Status: Offline
|
|
| We use HZ_1000 and NO_HZ, that should get quite a long way, but I can't promise that it will be enough for professional use (and no, we don't/ won't offer a dedicated -rt patched kernel). |
|
|
| |
|
|
|
|
|
finotti
|
|
|
Post subject: RE: low latency kernel
Posted: 26.01.2017, 14:30
|
|
Joined: 2010-09-12
Posts: 493
Status: Offline
|
|
Thanks for the reply.
Would it be relatively easy to compile my own version, based on your sources, to get a realtime version? I'd imagine one could create a patch, and when an updated comes, apply the patch to the source and compile it?
If so, would you have any recommended configurations. (Of course, I can look it up, I'd think, but I'd trust information coming from you more.)
I will try to see if there is any reason jack is choking (and giving xruns) besides the kernel... It would be much easier than always patching the kernel... |
|
|
| |
|
|
|
|
|
slh
|
|
|
Post subject: RE: low latency kernel
Posted: 26.01.2017, 23:15
|
|
Joined: 2010-08-25
Posts: 962
Status: Offline
|
|
| Given that -rt is a very invasive patchset, it definitely isn't "easy" - possible, sure, but not easy. It should be worth testing if the normal kernel, as configured for us, is already good enough for your use case, but as I don't really do anything more complex with audio than playing existing files, that's not really something I can comment upon (that unfortunately also goes for your further audio related enquiries, I simply don't need/ use ardour et al). |
|
|
| |
|
|
|
|
|
finotti
|
|
|
Post subject: Re: RE: low latency kernel
Posted: 27.01.2017, 16:31
|
|
Joined: 2010-09-12
Posts: 493
Status: Offline
|
|
|
slh wrote:
Given that -rt is a very invasive patchset, it definitely isn't "easy" - possible, sure, but not easy. It should be worth testing if the normal kernel, as configured for us, is already good enough for your use case, but as I don't really do anything more complex with audio than playing existing files, that's not really something I can comment upon (that unfortunately also goes for your further audio related enquiries, I simply don't need/ use ardour et al).
Thanks again for the info. I've been trying to "optimize" my system following this guide: http://wiki.linuxaudio.org/wiki/system_configuration. I am not sure how much of it is necessary (and how much of it is "snake oil"), but hopefully it will help.
In particular, that guide has some instructions on how to patch a kernel for RT (again, not sure how good it is, as I haven't even really looked at it carefully). I haven't tried it yet, but I have tried the RT kernel from Debian, and I am still having some overruns (unless latency is high), so the problem might not be the kernel. (I haven't compared both carefully to see if the RT one helps at all.)
I've been getting some help from https://linuxmusicians.com, so hopefully I will be able to pin point the problem.
In case anyone else might be interested, I might post here if, at the end of the day, the RT kernel makes a noticeable difference at all.
Again, thanks slh for the great support! |
|
|
| |
|
|
|
|
|
finotti
|
|
|
Post subject: RE: Re: RE: low latency kernel
Posted: 04.02.2017, 14:50
|
|
Joined: 2010-09-12
Posts: 493
Status: Offline
|
|
| FWIW, the problem I was having was due to the fact I was trying to use different cards for input and output. As soon as I set the same interface (Focusrite Scarlett 2i2) for both, I had no problems recording with very low latency (EDIT: 4ms, not 8ms) with the aptosid kernel. Just in case some was curious. |
Last edited by finotti on 08.02.2017, 22:27; edited 1 time in total
|
| |
|
|
|
|
|
slh
|
|
|
Post subject: RE: Re: RE: low latency kernel
Posted: 04.02.2017, 18:12
|
|
Joined: 2010-08-25
Posts: 962
Status: Offline
|
|
| Thanks for confirming that, quite a lot of -rt improvements have been merged into the normal kernel over the years (although -rt as a whole, including backmerging, has stalled in recent years), at the same time computers have also gotten much faster (and more importantly, multi-core SMP), so a dedicated -rt indeed shouldn't really be necessary in most cases anymore (unless you really require hard-rt, but higher level operating systems, like linux, aren't really the ideal solution for these anyways). |
|
|
| |
|
|
|
|
|
horo
|
|
|
Post subject: RE: Re: RE: low latency kernel
Posted: 24.02.2017, 21:36
|
|
Joined: 2010-09-11
Posts: 22
Location: Berlin & Lindau, Germany
Status: Offline
|
|
No need to patch a kernel on your own, why don't you use the debian unstable meta package linux-image-rt-amd64 that loads linux-image-4.9.0-2-rt-amd64 at the moment?
Ciao, Martin |
_________________
omnia vincit pecunia 
|
| |
|
|
|
|
|
finotti
|
|
|
Post subject: RE: Re: RE: low latency kernel
Posted: 24.02.2017, 22:15
|
|
Joined: 2010-09-12
Posts: 493
Status: Offline
|
|
| Thanks, horo, for the reply. I actually did that, but slh kernel is working just fine. I haven't tried pushing both to compare, but I've been able to get 4ms latency, which is fine for me. |
|
|
| |
|
|
|
|
|
|
