[LAU] irq sharing

[David Santamauro]

Hi,

On Tue, 24 Aug 2010 12:46:06 -0700
Niels Mayer <nielsmayer at gmail.com> wrote:

> On Tue, Aug 24, 2010 at 8:13 AM, David Santamauro <
> david.santamauro at gmail.com> wrote:
> 
> > thanks for the time. I only have one PCI slot, but 3 empty PCI-x
> > slots.
> >
> > I basically unplugged all USB devices as well as shut off both
> > network interfaces and on board audio interface in the bios and the
> > noise persists ...
> >
> > Not sure what to try next, this was a shot-in-the-dark.
> 
> 
> If you think it's a interrupt-sharing issue, consider using a
> pcie-to-pci riser to plug in your card. Apparently, this way of doing
> things may ensure less bus contention:
> http://old.nabble.com/does-a-pci-e-x1-to-pci-adapter-work-with-Linux-soundcards-and-existing-drivers--td29444687.html
> 
> 
> Also, be sure that interrupt contention is really the problem. If
> it's a constant noise, there's a better chance that it's a different
> problem. One issue is that your recording software may not be using
> the same bit depth, alignment, or complement as what the card
> expects. I can get this effect on output, for example, simply by
> using 'mplayer'

interesting angle ... 

envy24control (and the new mudita24) register this noise on input
whether devices are plugged in or not. So the noise is clearly coming
from either the delta breakout, card or driver. I've ruled out the
breakout box and card simply by the fact that the noise doesn't exist
on the same hardware with windows7 drivers.

There is no problem on any of the 10 outputs.


[ snip ]
 
> I'm thinking that something similar w/r/t bit
> order/depth/complement/etc might be happening on record/capture from
> your 1010?
> 
> What happens when you setup jackd to talk to your soundcard? My
> experience is that Jackd seems to do the right thing w/r/t matching
> the bit order/depth/complement expected by the soundcard.

jackd controls the delta 100% of the time.

> 
> -- Niels
> http://nielsmayer.com
> 
> PS: If this doesn't help, perhaps asking on the alsa-dev list may
> help? Alternately, perhaps someone from the ALSA project (Clemens
> Ladisch?)  may know about this problem and how to resolve it.

I will, thanks

David