Re: [Jack-Devel] xruns and server effects

PrevNext  Index
DateSun, 09 Nov 2014 08:20:36 -0500
From Paul Davis <[hidden] at linuxaudiosystems dot com>
ToMatt Flax <[hidden] at flatmax dot org>
CcJACK <[hidden] at lists dot jackaudio dot org>
In-Reply-ToMatt Flax [Jack-Devel] xruns and server effects
On Sun, Nov 9, 2014 at 7:11 AM, Matt Flax <[hidden]> wrote:

> Hi there,
>
> I was wondering for both Jack1 and Jack2 ...
>
> If a client xruns, does this effect the output to the underlying audio
> driver or are they autonomous ?
> If a client xruns, does this effect the other clients, or are they in
> their own threads ?
>


(1) xruns are not defined with respect to clients. It is the server that
detects an xrun using its own internal timing mechanism(s). An xrun means
that the entire JACK graph is late, and does not explicitly identify any
particular client. JACK1 resets the audio device when there is an xrun.

(2) since clients are autonomous processes, they do not share threads. But
in JACK2, the server may be "managing" clients from more than 1 thread. The
default mode of operation in JACK2 doesn't really care much about xruns, so
clients that take too long really don't affect anything else very much.
Using sync mode in JACK2 changes this. In JACK1 the server manages all
clients from a single thread.
PrevNext  Index

1415539244.15891_0.ltw:2, <CAFa_cKmiw-bJhYT=Dx+PG+=-jY86Yjjt7dgc6=sj5jpVDxgDjA at mail dot gmail dot com>