[Ardour-Users] Delta1010 + Edirol FA-101

[Ben Bell]

> The later is working. And with clock-synced sound-cards, alsa_in and alsa_out 
> don't need to resample at all...
> The hardware is also only a backend (but with the "I am hardware" flag set).
Hmm. A quick bit of digging makes me think we should probably be setting the
JackPortIsPhysical and JackPortIsTerminal flags within alsa_in and alsa_out,
then. Will try that.

> I did successfully run an presonus firepod, a focusrite pro24 and a
> development-board all clock-synced and aggregated into one large device. Of
> course, the latencies of the various in- and outputs are not the same as
> each device and io-type has a different converter latency. But as that is
> constant, its not a problem.

I hadn't thought of that. Presumably that's going to be an issue with alsa_in
as well. At a glance in the alsa_in source code there's nothing dealing with
reporting latencies in there so presumably any compensation would need to
be manual (including, perhaps, alignment of recorded material in ardour?).

Returning to FFADO -- if I had a pair of identical devices, is there
anything reason to expect that their latencies will be the same. Despite the
clock sync preventing drift, is they are still necessarily going to be
delivering their packets to the host server at different times, and perhaps
split at different points in the stream. Is this too small a thing to notice
or is there some need for compensation there too?

Perhaps it would be fine for two different instruments but not wise to split
a stereo channel between two devices.

Thanks for the answers so far,

bjb