[wdmaudiodev] Re: Modified kernel streaming example

  • From: User for realtime audio <audiodev@xxxxxxxxxxxxxxxxxx>
  • To: wdmaudiodev@xxxxxxxxxxxxx
  • Date: Thu, 04 Mar 2004 09:07:00 +0100

Well, actually not. I had this problem with one specific device only
and only for large buffers (2048 samples and more).

I have not followed this problem too much even though the technical support of the device
gave me a contact address for the driver I used. I did not contact the person however
because it seemed that with the latest version of the driver which is available in the meantime
the problem disappeared (no intensive testing!).

I modified my own software also to achieve better performance in another context.
Also this may be the reason why it seems to work now:

I am using the KS-example for capture and playback (duplex), not only for playback.
In the latest version of my software I have one thread for incoming and one for outgoing buffers.
Before I had one thread for both, waiting for all events in one Wait-function.

Mmmh, the problem I described was only for playback... Maybe it is helpful anyway.

But there was no change in the context of KS-Interfacing actually...

Best Regards,
Hauke

Jerry Evans wrote:

I have run into exactly the same problem with the AC97 drivers in WinXP SP1.
Did you find a solution ?

Does anyone know what to do when this occurs ?

TIA

Jerry

----- Original Message ----- From: "User for realtime audio" <audiodev@xxxxxxxxxxxxxxxxxx>
To: <wdmaudiodev@xxxxxxxxxxxxx>
Sent: Wednesday, December 17, 2003 11:20 AM
Subject: [wdmaudiodev] Re: Modified kernel streaming example


Well, still suffering the same problem I have found out something new:

It seems that my problem is correlated to the order of the buffers being
processed
by the driver:
If using 8 Buffers to flush the order to write the buffers should be
0..7 according to the
first write-accesses to the driver. And that should be the order of the
signalled events.

Whenever using buffers with critical sizes the order in which the
buffer-writes have been
completed are kind of arbitrary, e.g. 0 5 3 7 0 2 5 6.

In case of those buffersizes that function propperly, the order is as
expected 0..7.
Thus it seems that the driver somehow writes the buffers to the hardware
in a wrong order, thus my question:



******************

WDMAUDIODEV addresses:
Post message: mailto:wdmaudiodev@xxxxxxxxxxxxx
Subscribe:    mailto:wdmaudiodev-request@xxxxxxxxxxxxx?subject=subscribe
Unsubscribe:  mailto:wdmaudiodev-request@xxxxxxxxxxxxx?subject=unsubscribe
Moderator:    mailto:wdmaudiodev-moderators@xxxxxxxxxxxxx

URL to WDMAUDIODEV page:
http://www.wdmaudiodev.de/






******************

WDMAUDIODEV addresses:
Post message: mailto:wdmaudiodev@xxxxxxxxxxxxx
Subscribe:    mailto:wdmaudiodev-request@xxxxxxxxxxxxx?subject=subscribe
Unsubscribe:  mailto:wdmaudiodev-request@xxxxxxxxxxxxx?subject=unsubscribe
Moderator:    mailto:wdmaudiodev-moderators@xxxxxxxxxxxxx

URL to WDMAUDIODEV page:
http://www.wdmaudiodev.de/

Other related posts:

  1. Home
  2. wdmaudiodev
  3. Archive
  4. 03-2004