Re: [Ekiga-list] A comparison ALSA-PULSE

[Damien Sandras <dsandras seconix com>]

Le samedi 21 février 2009 à 18:29 +0000, Andrea a écrit :
> Hi,
> 
> I've tried to compare the behaviour of ALSA when Ekiga uses the direct access vs going through
> pulse-alsa module.
> 
> I've added some alsa print status in ptlib and the 2 files are the output when playing the ring-tone.
> 
> I've used the 2 alsa calls to populate the log
> 
> 1) snd_pcm_dump at the beginning
> 2) snd_pcm_status_dump after each write
> 
> The first part is the same, but the status after each write is different
> 
> DIRECT
> 
> About to write 1764 len bytes
>   state       : RUNNING
>   trigger_time: 9762.542972003	<<<<<<<<<<<<<<<<<<<<<<<<
>   tstamp      : 9762.543004130  <<<<<<<<<<<<<<<<<<<<<<<<
>   delay       : 430		<<<<<<<<<<<<<<<<<<<<<<<
>   avail       : 1334
>   avail_max   : 1334
> 
> PULSE
> 
> About to write 1764 len bytes
>   state       : RUNNING
>   trigger_time: 1235218239.552437000	<<<<<<<<<<<<<<<
>   tstamp      : 0.000000		<<<<<<<<<<<<<<<
>   delay       : 0			<<<<<<<<<<<<<<<
>   avail       : 441
>   avail_max   : 1764
> 
> 
> And then every now and then when running via pulse we generate an underrun
> 
> About to write 1764 len bytes
>   state       : RUNNING
>   trigger_time: 1235218239.552437000
>   tstamp      : 0.000000
>   delay       : 0
>   avail       : 0
>   avail_max   : 1764
> #######################################################################  EPIPE
>   state       : XRUN
>   trigger_time: 1235218239.552437000
>   tstamp      : 0.000000
>   delay       : 0
>   avail       : 0
>   avail_max   : 1764
> 
> 
> Does anybody know a bit what all those values mean?
> It seems that the underrun comes earlier that without pulse-alsa...
> 
> I've tried to increase the size of the buffers in PSoundChannelALSA::SetBuffers() and the playback
> is much better, but that gives a bigger latency and delay when calling the echo test.

The pulseaudio guy told us we have a problem with the way we calculate
the periods. Not sure exactly what is wrong though...
-- 
 _     Damien Sandras
(o-      
//\    Ekiga Softphone : http://www.ekiga.org/
v_/_   Be IP           : http://www.beip.be/
       FOSDEM          : http://www.fosdem.org/
       SIP Phone       : sip:dsandras ekiga net