neod and pusleaudio -- evil tandem? ;-)
sears at cs.berkeley.edu
Sat Jul 26 00:11:13 CEST 2008
I'm trying to improve mediaplayer, and a big part of that is improving
performance. If you figure out why pulseaudio eats 6% CPU, that'd be
great! Also, it mysteriously eats much more than that when music is
being played back.
My theory is that it's doing unnecessary signal resampling and
int->float->int conversions. Having mediaplayer bypass pulseaudio and
talk directly to alsa saves a lot of CPU.
It's probably some sort of misconfiguration. This bug might be relevant:
If neod is using gstreamer, switching to ALSA might be trivial. It was
a one line patch to switch mediaplayer from Pulse to ALSA.
Yaroslav Halchenko wrote:
> ok -- missing sound... it seems that pulseaudio doesn't loose any socket
> connection after resume -- it still gets those inputs from the screen,
> lsof is the same prior and post suspend, but no sound.
> also according to dumps of alsactl before and after resume nothing
> changes and they correspond to stereoout.state
> where else to look?
>> and then I have the same problem with loosing sound after resume -- (I
>> simply have dim/lock PM strategy on) -- pulseaudio goes busy for a few
>> seconds after resume, and then sounds are gone (no tapping sound no
>> ringing sound -- just vibration). And that of cause removes pulseaudio
>> from the list of 'busy' processes although pulseaudio is still running
>> (neod remains as busy as before 9-10% CPU). Restart of pulseaudio and
>> then x-server (to have dialer/today reconnected to pulseaudio) helps
>> until next suspend (and they both again run at those 16% CPU ;-))
>> So -- any ideas?
More information about the support