Re: [Jack-Devel] is this latency coming from ALSA, jack, or ardour?

"Chris Caudle" <[hidden]> writes:

> I wanted to test out the new and better latency reporting in current jack,
> so I made a session in Ardour 3 Alpha 4, turned on the click track and
> then made the following three sessions:
>
> jack 1.9.6, click track routed to track 1. Stopped transport, armed track 2
> turned off click, routed track 1 to track 2. Stopped transport, armed track 3
> routed track 2 to track 3. stopped transport, armed track 4,
> etc.
>
> So all internal jack routing, no audio interface involved.  As expected,
> track 10 lined up perfectly with track 1.
>
> So then I used my USB audio interface (Lexicon Lambda) with Alpha 4 and
> jack 1.9.6.  By the time I got to track 10 with bounces, there was over 1
> second of difference between track 1 and track 10.  Not really usable for
> anything with lots of bounces.
>
> I deleted all the jack 1.9.6 files, compiled and installed jack 2 from svn
> (jackmp revision 4293), and did the same thing.  Now the click to track 1
> offset is around 13ms, and the track 1 to track 9 offset is about 103 ms. 
> I had the levels set too low, so I couldn't really see the onset of attack
> by track 10 well enough to measure, but I assume bouncing to track 10
> would add about 13ms more, just like the original click to track 1 bounce.
>
> Is that as good as I can expect with a USB interface, or is jack 0.120
> better than that?
> I have a PCI interface available, so should I try that to compare against
> the USB interface?  It took me about an hour to do this much, so it would
> be tomorrow night before I could get back to it.  Just wondering whether
> it is a useful exercise at all, and if it might be, which would be the
> more useful info, jack2 SVN compared to jack 0.120, or USB interface
> compared to PCI interface both using jack2 SVN.  Now that I think about
> it, I suppose comparing the PCI to the USB would at least show whether the
> interface was just reporting an incorrect latency value, which would
> exonerate jack2 from the problem and point to ALSA if the same bounce test
> worked with the PCI card.

Do you get better results with jack2 in sync mode? jack2 default mode is
async. AFAIK jack2 in sync mode should give you same latency as jack1.

-- 
Nedko Arnaudov <GnuPG KeyID: 5D1B58ED>

1302709200.459_0.ltw:2,a <87pqoq5fdg.fsf at usbix dot spacelabs dot org>