On 03/21/2012 12:54 PM, James Warden wrote:
Hi Fons,
I gave it a shot yesterday evening. It was not meant to be a thorough
test but just a check whether I could just replace alsa_in/out with
zita_a2j/j2a in my environment:
- Core 2 Duo 2 x 2.4GHz (cpufreq disabled)
- kernel 3.2 from liquorix (debian i686) with thread_irq enabled (but
no complete preemption patch) - RME HDSP + Multiface II
- jackd2 (latest from svn, before it got moved to github) running in
sync mode, at 96kHz, buffer size 2 x 256 frames
- zita_a2j running on ALSA loopback (loopback card with only two pcm
substreams enabled when loaded to the kernel)
My .asoundrc is described in the unofficial ALSA WIKI page about the
loopback device used as a bridge for non-jack apps. I basically set
the loopback card h/w and s/w parameters to match my jack setting.
The result is not as good out of the box as with alsa_in. I have a
wav file which is a 440Hz signal running for 2mn. When I play it
through mplayer -ao alsa and alsa_in on the other end of the ALSA
loopback, I get no interruption at all. Your zita_a2j app quickly
introduces interruptions (dropouts probably) which happen when it
displays something about synchronization in the terminal. I played
with the resampling factor but it did not improve anything in terms
of dropouts. On the other hand, alsa_in was just fine.
Not sure whether you can use this at all as a data point, but
something is not optimally running. I have no time to do more
thorough tests.
I've been debugging this use-case with Fons over the last 24 hours.
Meanwhile he's at zita-ajbridge 0.2.1 (still unreleased) and it
currently runs [mostly] find as a drop-in replacement for alsa_in/out
[for loopback devices].
There's still some issues to be sorted out though.
robin