18 Nov
2005
18 Nov
'05
7:33 a.m.
Cesare Marilungo wrote:
Clemens Ladisch wrote:
I still have some doubts. 24bits at 48 kHz full duplex in stereo is
4.8Mb/s which is less than half the total theoretical bandwidth. But
we're talking about something that must be realtime without loosing
anything. How does latency relate to this? I'm not much into such
technical aspects but I believe that with more bandwidth and less
latency you need more processor time and speed. So, pratically with
current hw you can't go too far with usb 1.0. I achieved 5ms total
latency but the system was unstable, on a 2GHz AMD64.
Please correct me if I'm wrong. I just want to learn.
Anyway do you have experience with this Audigy NX? What about the
latency? My brother has an older Audigy (PCI not USB) and with the
original drivers he get 1sec of latency (300ms in + 700out).
Cesare Marilungo wrote:
Is it true? I didn't know.
I can run it at 11.6ms (5.8+5.8) full duplex
without any xrun. Lower
latencies work but I get some xrun every now and then.
All USB audio devices move data in one-millisecond packets, so the
achievable
latency depends _only_ on the software, not on what device you use.
Anyway I
doubt you can get a greater bandwith with usb 1 at 24bit.
The theoretical limit of USB 1.x is 12 Mb/s (i.e., 1,500,000 bytes/s).
The Audigy 2 NX can do eight channels with 16 bits at 48 kHz (750 KB/s),
but full duplex stereo with 24 bits at 96 kHz would need 1125 KB/s which
is too much considering the protocol overhead.
Regards,
Clemens