[LAU] Small instrument hardware module
len at ovenwerks.net
Wed Oct 22 23:47:05 UTC 2014
On Wed, 22 Oct 2014, Kazakore wrote:
> But yeah there is a ~250 page pdf on the v1 docs which can be found online
> (in fact it's linked from the Wikipedia article I linked to earlier.) Not
> read through it myself. But from the desctription I can understand why you
> say Bus. But a bus wouldn't generally include codecs and things, so it's more
> than just the bus really isn't it? Nearly a whole protocol really...
I don't think the codecs themselves are a part of it. But the way they
talk to it is, and so most codecs are "made for HDA" chips. I think any
codec can be used with the right glue.... just like with a PCI(e) AI the
glue has to work with the PCI bus.
The main thing though, is that HDA audio is not a name of an audio
interface that is specific, so much as a group of similarily accessed AIs.
I think the bus structure or protocol does limit how low latency can go. I
am not sure if running higher sample rates (96/192k) would give lower
latency or not as the transport speed is still 48k in any case. However
filling up the buffer faster may give lower latency. I think the HDA spec
is designed to have the same latency as the rest of the intel system with
no special tweaks. That is, still using all the selling point features,
such as "Boost technology" or "Hyper threading", both of which increase
"Pros" are expected to be rich and willing to spend money on special
interfaces and computing engines, not try to get $10k plus ($100k is not a
lot for some of this stuff) performance from $1k minus studios :) Though
I can't see as that has ever stopped anyone. My point though is that what
we are doing is not what Intel has designed desktop CPUs and related chip
sets for. It is something we do in spite of the design.
More information about the Linux-audio-user