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

On Fri, April 15, 2011 1:22 pm, Paul Davis wrote:
> you've got a digital loopback, it appears.

How so?  I have a cable connecting the stereo analog outputs of my card
back to the stereo analog inputs of the card.
Oh, wait a minute.  <ears turn red>
In the connection panel on qjackctl, I connected jack_delay:out to
jack_delay:in.
That wasn't right, was it?  Obvious now that I think about it, given how
jack presents connections.

OK, so now I connected jack_delay out to the card physical output port,
then connected the card physical input port to jack_delay in, and got
this:

4155.593 frames     94.231 ms

That makes much more sense.

Assuming that precision is correct, does jack take fractional frames as
part of the argument to -I and -O?  Qjackctl does not allow fractional
frames to be input through the dialog.
Wait, 1 frame = 1 sample period? Never mind, who cares, I was confused
with periods.

> my instructions were wrong. sorry. subtract frames/period from
> jack_iodelay's answer, then divide by 2 for the input/output numbers.

OK, with your correction, and my corrected connection, that makes a lot
more sense:
(4155.593-3072)/2 = 541.79 frames additional latency,  in and out.

> note that dividing by 2 is a heuristic: if you used different devices
> for A/D and D/A, it would not necessarily be correct.

Yes, that is understandable, it isn't possible to distinguish input path
from output path effects when you have just one loopback connection.

It should in principle be possible to calculate the right values if you
happen to have a device with really thorough specs from the designer.  I
know Prism specs the delay through the AD and DA sides of their
converters, digital in to analog out and analog in to digital out.  The
two sides are different by about 2x.

Trying to figure this out has given me new appreciation for why the
network audio guys wanted to time stamp everything and just line all the
audio up via time stamp, rather than trying to figure out path latency.

These measurements were on the PCI card.
I will have to apply this now to the USB interface, and see if that
corrects the somewhat unreasonable behavior.

Thanks for the help, this is very useful.

-- 
Chris Caudle

1302894271.26882_0.ltw:2,a <7923c9f8ef1361af9bee81a2189ebe9d.squirrel at email dot powweb dot com>