Re: [Jack-Devel] Interpreting 'Clients end date' gnuplot diagram

The audio interrupt indeed seems suspicious.  Can you put other diagrams on your site?

You can also try jackd in "synchronous" mode like:

jackd -S -P83 -dalsa -dhw:0 -r48000 -p64 -n3 -Xraw

Stéphane


Le 18 mars 2012 à 15:42, Dénes Almási a écrit :

> Sorry, I thought it is a well-known form of profiling. This
> method was earlier suggested for me by Stéphane Letz.
> 
> I compiled jack with --profile, it is a diagram generated by the
> server itself. Then I used the autogenerated generate_timings
> script to plot the diagrams. This one is explained in
> http://www.grame.fr/Ressources/pub/Timing.pdf , section 2.3.
> 
> I was asking this because it is difficult to understand these
> results at first glance even with the explanations in the paper.
> 
> Measuring was done by executing,
> $ jackd -P83 -dalsa -dhw:0 -r48000 -p64 -n3 -Xraw
> then started my app, waited for some xruns to happen
> and closed everything. When the server closed, it generated
> this profiling output among others.
> 
> I just want to understand this whole architecture in order
> to be able to create efficient realtime applications.
> 
> Thanks,
> Dennis
> 
> On 2012-03-18 15:31, Fons Adriaensen wrote:
>> On Sun, Mar 18, 2012 at 03:09:26PM +0100, Dénes Almási wrote:
>> 
>>> Just to be sure, I am asking: Does the following
>>> diagram mean that its not my app who is late but the
>>> audio interrupts are not regular enough?
>>> 
>>> http://denes.rudanium.org/shots/perform-clientsenddate.png
>> 
>> Difficult to say if you don't even explain what this data is
>> and how it was measured.
> 
> Jack-Devel mailing list
> [hidden]
> http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

1332082825.24704_0.ltw:2,a <25DC3E3C-D4CE-4038-A7F5-2E5943F9FF5B at grame dot fr>