[pulseaudio-discuss] Underruns with mpd+pavucontrol

[Jan Kasprzak]

	Hello,

I use PulseAudio with mpd (www.musicpd.org), and I experience underruns (skips)
every time I switch virtual desktops or do anything graphically intesive.
I have a dual-core CPU, and as far as I can see (in top(1) or in
system-monitor-applet in GNOME) the CPU is not too busy: when switching
virtual desktops, I can see load spikes of about 130 % of a single CPU
in system-monitor-applet (0.5 s measurement interval), but I think
mpd and/or pulse should have buffers big enough to overcome this.
Moreover, the CPU load is not even big enough for ondemand cpufreq
governor to increase the frequency.

	I have discovered that underruns go away when I don't have
pavucontrol running. My guess is that pulseaudio sends the data for
the volume meter to pavucontrol, and when pavucontrol is not able to
read the data (because it is blocked on sending commands to the X server,
which is in turn busy unmapping/remapping windows during a virtual desktop
switch), the pulseaudio daemon also blocks and underrun occurs.

	So maybe communication with pavucontrol should be made
more "non-blocking"? Altough it apparently _is_ non-blocking to some
extent - when I suspend pavucontrol using Ctrl-Z, there is no underrun.

	Thanks for any hints how to fix this. I need pavucontrol running
for switching the player between the speakers sink and headpohones sink.
My system is Fedora 10/x86-64, pulseaudio-0.9.13, pavucontrol-0.9.7, mpd-0.14.

-Yenya


-- 
| Jan "Yenya" Kasprzak  <kas at {fi.muni.cz - work | yenya.net - private}> |
| GPG: ID 1024/D3498839      Fingerprint 0D99A7FB206605D7 8B35FCDE05B18A5E |
| http://www.fi.muni.cz/~kas/    Journal: http://www.fi.muni.cz/~kas/blog/ |
>>  If you find yourself arguing with Alan Cox, you’re _probably_ wrong.  <<
>>     --James Morris in "How and Why You Should Become a Kernel Hacker"  <<