[wdmaudiodev] Re: WaveRT peculiarities

  • From: Robert Bielik <robert.bielik@xxxxxxxxxx>
  • To: "wdmaudiodev@xxxxxxxxxxxxx" <wdmaudiodev@xxxxxxxxxxxxx>
  • Date: Mon, 19 Jul 2010 10:35:32 +0200

Robert Bielik skrev:
Why would a lower latency setting work better ? Scheduling granularity ? More optimized buffer size for the driver ? The driver always returns my requested buffer size, regardless of how odd it might be...

Hehe... it seems that at 150 frames buffer, the pops/clicks were "masked" by my 
testsignal, changing the test signal
revealed the pops/clicks at 150 frames also. But now I can stream easily with 
128 frames, even 64 frames latency by
working out how to use the hw latency parameter from the driver (in conjunction 
with position register).

But it seems odd that in the render case, I need to add hw latency to position register to get the last frame read by the hw FIFO, but in the capture case, if I do that (or subtract the hw latency) the resulting position is "off". Thus
in capture case, I don't use the hw latency parameter at all, which makes it 
work fine (I use buffer with 2 notifications
per buffer, and with proper adjustments the resulting position is pretty much 
exactly at zero and half buffer offset).

Ideas why this might be ? Acc. to the WaveRT document, AFAIK the hw latency 
should be used in the capture case also ?


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


Other related posts: