On Mon, Dec 31, 2012 at 5:32 PM, Len Ovens <len@ovenwerks.net> wrote:

On Mon, December 31, 2012 6:58 am, Paul Davis wrote:
> On Mon, Dec 31, 2012 at 9:47 AM, Len Ovens <len@ovenwerks.net> wrote:
>
>>
>> On Sun, December 30, 2012 7:12 pm, Devin Anderson wrote:
>>
>> >
>> > JACK *could* deal with this issue, but:
>> >
>> > 1.) It would be better if the functionality was independent of JACK.
>> > 2.) I don't think the issue is JACK's responsibility.
>>
>> Quite honestly, it does not seem to be something that _should_ be done
>> by
>> ALSA either, because properly joining two devices would mean an extra
>> buffer (and latency).
>
>
> this is not true, if they are clock-synced.

Well there is sync and sync. It seems the cards are in "sync" in that they
are exactly the same frequency, but not in "sync" phase wise in the
transfer of data.

"startup" of linked PCM devices should take less than a handful of milliseconds. if it doesn't then i suggest that its a driver-specific issue, not an ALSA one. the internal driver design has this in mind. if they are <<1 sample "out of phase" it should not make any difference *if* ALSA waits for the last/slowest device.