rt kernel

[Ralf Mardorf]

On Wed, 2011-04-06 at 14:40 +0200, ailo wrote:
> On 04/06/2011 12:54 PM, Ralf Mardorf wrote:
> >> How have you tested? What results have you obtained? If you want
> >> convince someone else you should provide testing procedures and
> >> numeric results.
> > 
> > If I play a MIDI track with external MIDI equipment, DX7, Matrix-1000,
> > D4, TG33 and others with 1000 Hz timer and record it to an audio track
> > and watch the files by Audacity, I can see that there's more jitter,
> > than when using the HR timer.
> > I also recorded kick, hi hat and snare one after the other and the
> > rhythm didn't groove anymore.
> > On LAD, 64 Studio lists, perhaps JACK devel and others too I posted
> > tests years ago. All regarding to this issue and other sync issues,
> > because Linux lacks on sync.
> > 
> 
> How recently have you tested -generic and -lowlatency?
> 
> 
> -- 
> ailo
> 

I didn't really tested 'lowlatency', because even the RT-patch doesn't
fit to my needs, hence I suspect that preemption only wouldn't be ok,
even if it should be as good as preemption rt.

I once used a preemption only kernel, had some issues, asked for help at
JACK devel list or another list and they asked me to use a kernel
preemption + rt. IIRC the issue had nothing to do with the preemption
only kernel, it still was there when using a preemption + rt kernel,
anyway, if you report async issues and xruns and stuff like that for
jackd, resp. jackd clients and you aren't using a kernel-rt, the first
hint you'll get, is to use a kernel-rt and depending to the kind of
problem, nobody will take a look to your issues, before you do this.

Default and vanilla kernels don't do what I do need. For this I don't
need to do tests, resp. launching an audio app and just moving the ps2
mouse will demonstrate the limitations of a default kernel(-desktop,
-generic or what ever a distro does call the default kernel).

Greetz,

Ralf