pulseaudio excessively drains laptop battery

[Maciej Borzecki]

Binary package hint: pulseaudio

After installing latest pulseaudio package and staring the server the power consumption raised up to 24W on laptop. Having stopped daemon it dropped back to ~15W. Remarkably, this behaviour was observed without any music being played. This behaviour is unacceptable and prevents from using pulseaudio on laptop.

[Joe Baker (joebaker)]

Could it be that you were transcoding one sample rate to another?

maybe if there was unity between all the devices the load would be less.

[Maciej Borzecki (maciek-borzecki)]

Yes, but the load was observed without any sound being played, thus nothing was being resampled. For me it seems as if even when no real sound samples are sent to the card, there is some activity which contributes largely to draining the battery.

[Joe Baker (joebaker)]

maciek wrote the following on 09/14/2007 01:39 AM:
> Yes, but the load was observed without any sound being played, thus
> nothing was being resampled. For me it seems as if even when no real
> sound samples are sent to the card, there is some activity which
> contributes largely to draining the battery.
>
>
I can imagine that silence would be re-sampled just as any sound would
be. Unless there is some sort of intelligent muting circuit. It was a
good call to see that this was draining laptop batteries though.

-Joe Baker

[Jochen Kemnade (jochenkemnade)]

A fix for that issue was committed to pulseaudio SVN and will probably be contained in version 0.9.7.
See pulseaudio ticket 114: http://www.pulseaudio.org/ticket/114

[diegoe (diegoe-deactivatedaccount-deactivatedaccount)]

Using powertop you can confirm that pulseaudio 0.9.6 raises wakeups for the audio chip. IIRC .7 included among other things fixes for device polling.

[zdzichu (zdzichu-gmail)]

Can we get PA 0.9.7 backport to Gutsy? Debian package entered testing on Wed, 31 Oct 2007

[Daniel T Chen (crimsun)]

Please see https://help.ubuntu.com/community/UbuntuBackports#head-37a793d5ee480081f1c9f19e07fcdcdae5e6a9ed

[Daniel T Chen (crimsun)]

(Sorry, issue is fixed upstream, but the latest comment needs a backport reference.)