8 Jan
2012
8 Jan
'12
3:12 p.m.
Chris Metzler wrote:
No, two channels. The limits are 512000 bytes/s per direction full
duplex or 1023000 bytes/s half duplex.
Mono would be possible, but I don't know of any such device.
Regards,
Clemens
Yeah, this was one of the questions I meant to ask,
and forgot --
whether there are PCIe options.
AFAIK the only PCIe sound chips are made by Creative (consumer
cards, no good Linux support). It is possible to put a PCIe/PCI
bridge chip on a card, but the only vendors that do are ESI and
Asus. ESI uses Tenor chips, which aren't supported at all; Asus
Xonar cards are useful only if you do not need more than one stereo
consumer-level line input (but then they have very good ADCs; see
<http://www.alsa-project.org/~clemens/xonar-models.html>).
Clemens Ladisch <clemens(a)ladisch.de> wrote:
Maybe. Usually, there are just too many onboard devices.
Chris Metzler wrote:
Good point. But they don't typically share those IRQs, do they? Or if
they do, isn't it with a PCI slot, so that you might benefit by moving
anything in that slot? - Latency tuning possibly required (identifying
IRQs associated with
PCI slots and picking slot accordingly)
USB and 1234 controllers are PCI devices too, so you have exactly the
same IRQ problems, except that you cannot change the slot of an
onboard device. USB has lots
of overhead. It can just barely fit eight 16/48 channels
into USB 1.1; full-duplex 24/96 is not possible.
Just to make sure I understand what you're saying here: 24/96 means
2.3Mbps, so full duplex would mean 4.6Mbps, which is less than the
USB1.1 12Mbps hard limit. But there's so much overhead in USB traffic
that you don't really ever come close to having 12Mbps available in
USB1.1, and in fact wouldn't have 4.6MBps available for one channel
24/96 full duplex?