Re: [Jack-Devel] Place for bug reporting
Hi Chris, thanks for reply!
I have a 720MHz ARM CPU which have only one heavy task (it's jackd).
Also I patch my kernel to PREEMPT_RT support, change governor to
performance, change linux interrupt frequency to 1000HZ (SETUP_HZ_1000
and SETUP_HZ=1000) and recheck CPU frequency (720Mhz). Do you think this
hardware is not suitable my scheme? With my simple network clients
(before xruns occur) I have CPU load about 4-5%, but I didn't solve the
synchronisation task yet, so both ends have own synchronisation (this
task will be done by jitter buffer and voice detector) and the same
sample rate. But as I said earlier after software got xrun loop CPU
became highly loaded so even console is not working properly and I
forced to reset board. Thanks for your tip I will check sound generator
with netone and return back.
04.06.2014 1:15, Chris Caudle пишет:
> On Tue, June 3, 2014 8:28 am, TimKa wrote:
> [ description of network setup which resulted in overruns, including
> messages:]
>
> JackEngine::XRun: client = alsa_in was not finished, state = Triggered
> JackEngine::XRun: client = alsa_out was not finished, state = Triggered
>
> The alsa_in and alsa_out clients perform real time resampling in order to
> accomodate the sampling clock differences between the netjack master and
> the local hardware sampling clock. The error messages show that alsa_in
> and alsa_out are causing the xruns, so likely the processor you have does
> not have the performance needed for real time resampling, or there is
> something in the alsa_in and alsa_out code which does not compile well on
> ARM 32 bit.
>
> You could verify that by removing the need for alsa_in and alsa_out by
> generating audio on the netjack slave rather than using the audio
> hardware, either synthesize audio signals or play back a pre-recorded
> file.
>
1401867936.1350_0.ltw:2,a <538ECE76.4050805 at yandex dot ru>