Yes, syncing separate soundcards is pretty well impossible unless one of them can be slaved to the word clock of the other (as some cards do offer, indeed).
However, the best solution for this task is a true multi-channel card, such as the Sonorus STUDI/O, which can be addressed as, for example, one eight-channel device. The Sonorus does have Linux support, including the 8-channel mode (through the ADAT lightpipe, so you would need a converter). There are several such cards available for Windows, of course (see my Soundcard Attrition Page for details - URL below).
The consequence of this is that only one device needs to be opened; you would then write a quad stream just as easily as one writes a stereo one. Needless to say, support for such a device would need to be incorporated into the PD code. It's on my list of "things I want to do" for PD under Windows - in my case, driving the Creamware Pulsar running its '16-channel Interleaved' mode. Creamware are promising Linux support sometime, but certainly not soon, and Very Unlikely to be Open Source.
Richard Dobson
Karl MacMillan wrote:
Marius
I think that there are some sync issues when using two sound cards in this way that may only show up when trying to do real time processing like PD. As far as I know there is no provision for this in PD at the moment but I have written the beginnings of this into the ALSA driver code that is in the latest release. One of the questions is how to specify which devices to open on the command line. Any thoughts anyone? The -soundoutdev and -soundindev only apply to the RME Hammerfall card.
Karl
s_medge wrote:
hi,
some programs (w95) can handle two soundcards simultanously. i actually run pd with different devices for soundin and soundout. (no full duplex cards). my question is, if it would be possible to use both sounddevices for output (four channels)? is there an argument like soundoutdev 1+2???
marius.