On Sat, Jan 01, 2011 at 05:35:14PM +0100, Jörn
Nettingsmeier wrote:
On 12/31/2010 10:03 PM, Philipp Überbacher
wrote:
Excerpts from torbenh's message of 2010-12-30
16:30:33 +0100:
>>- smp aware
>>- backendswitching
Means for a user?
that a suitably parallel jack graph will use more than one cpu core,
which is a major win.
backend switching means you can change the backend at runtime
without having to tear down your jack graph, which is also very
practical in some situations. (haven't tried it yet, but iiuc you
can move from ffado to alsa backend or between different
alsa-supported cards.)
yes. thats right.
you can also change the samplerate, but thats not very well supported
yet.
another nice thing is that you can free the soundcard by switching to
the dummy driver. jack graph still runs. soundcard can be used by some
other software.
>>>>- strictly synchronous like jack1. (-> no latency penalty)
>
>Async (jack2?) has a latency penalty?
yes.
when you want clickless connections, yes, afaik.
from jack2 JackAlsaDriver.cpp:
// Add one buffer more latency if "async" mode is used...
port->SetLatency((alsa_driver->frames_per_cycle *
(alsa_driver->user_nperiods - 1)) +
((fEngineControl->fSyncMode) ? 0 :
fEngineControl->fBufferSize) + alsa_driver->playback_frame_latency);
>>-
clickless connections.
>>- shuts down audio processing when cpu is overloaded for too long.
I guess most users don't know very much about the technical details of
the jack implementations.
which is good in principle (since it seems to just work), but kind
of implies we are lacking documentation. check out the jackaudio
wiki, torben in particular has been very active there. the wiki
seems to be known little, but there's a link from the jackaudio
front page.
i doubt, that we documented all the technical details though.
Thanks everyone, that pretty much clears it up for me. Seems like
tschack is the most advanced jack implementation now in many ways. It
did already work well when I tried it some time ago, I guess it's time
to try it again
Thanks,
Philipp