Hi, On Thu, Feb 24, 2011 at 10:14 AM, Cristina Cristea <doina.cristina at gmail.com> wrote: > On 02/23/2011 04:08 PM, Maarten Bosmans wrote: >> >> Please keep the conversation on the mailing list. (though it looks >> like I forgot to do that too) >> >> 2011/2/23 Cristina Cristea<doina.cristina at gmail.com>: >> >>> >>> No, the glitches are now more rarely and smaller. But, after 31 minutes >>> of >>> playing (with only one small glitch, bearly audible, on min 27), >>> PulseAudio >>> started to display a lot of underrun messages and died. >>> >> >> Died, as in crashed? Verbose log output would be interesting here. >> Most likely the CPU load limiter kicked in and pulse shut itself down. >> > > I don't have the CPU load limiter activated in my daemon.conf and there is > not much information in PulseAudio log. > ....... > I: module-loopback.c: Coud not peek into queue > I: module-loopback.c: Coud not peek into queue > I: module-loopback.c: Coud not peek into queue > Killed > After playing for a while with only small glitches and a constant CPU usage > of around 80%, the ?pulseaudio CPU usage reaches over 100%, bigger glitches > are showing and PulseAudio is killed. > >> >>> >>> Also, module-loopback causes PulseAudio to be a a very big CPU consumer >>> on >>> my system-like about 80%. >>> >> >> That should not be. Can you get this behaviour also without >> module-loopback? (e.g. by recording from the bluetooth source with >> parec) For further information on how to debug this, see >> http://pulseaudio.org/wiki/HowToDebugCPULoadBugs >> >> > > For recording from bluetooth source with parec, CPU usage is around: > 5%- lt-pulseaudio > 2% -lt-pacat > > When playing the recorded file, with paplay: > 55% - ?lt-pulseaudio > 5%-lt-pacat > Btw, what system are you testing this? It seems the sbc decoder is causing quite a heavy load in your system. -- Luiz Augusto von Dentz Computer Engineer
