On Thu, Apr 06, 2006 at 11:39:14PM +0300, Jan Knutar wrote:
> On Thursday 06 April 2006 19:26, Jack Carroll wrote:
>
> > Yup, I agree. It takes non-zero time to convert the raw bit stream
> > into Ethernet frames. One feature the protocol would have to support is
> > synchronization of the streams coming from different digitizer channels,
> > including digitizers on different subnets. I'd say 10 microsecond
> > synchonization accuracy would probably be adequate; that's about 0.1" or 2.5
> > mm sound source location accuracy.
>
> Would maybe be a good idea to ask the Jack people about latencies. I've seen
> people on LKML who've tried to do audio with latencies that I thought were
> absolutely insane at first.
> I'm not convinced ethernet is quick enough :)
There are a couple of ways around that. Use ATM instead of TCP/IP
over Ethernet. Or include accurate time stamps in the data stream. Skew
over the network isn't important, as long as the software can match up the
data on the disk from the different input channels.
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
