[MPlayer-dev-eng] A/V Sync on SMP

Attila Kinali attila at kinali.ch
Sat Mar 6 13:35:00 CET 2010


On Sat, 6 Mar 2010 13:31:39 +1300
Keith Chew <keith.chew at gmail.com> wrote:

> First, if the mplayer-users mailing-list is more appropriate for this
> question, please advise. 

Overall it looks like a user question/problem, so -users would
be the apropriate place.

>I am posting this here because I found an old
> post on this list "A-V sync tuneup, round 2", which relates to the
> problem I am observing.

You are refering here to a thread from november 2001.
ie a time very early in MPlayer development. I'm not sure
whether anything said there still applies to the A/V code.
 
> When the kernel is compiled without SMP support, I can play a video
> with 6 channel wav audio (using ALSA as -ao) without any problems. To
> check, I ran 4 instances of mplayer playing the video simultaneously,
> making the CPU hit 80%. All fine.
> 
> When the kernel is compiled with SMP (chip is Intel Core 2 CPU T5600),
> 1 instance of mplayer fails to play the video, giving the "Your system
> is too SLOW to play this!" error within 10 seconds of playback.

This can be caused by anything. Starting from broken sound card
driver to a misconfigured xorg. Please read bugreports.html
and provide all informations outlined there.
> 
> Initially, I thought it was a kernel problem, so I tried many
> different compile options, eg increase Hz to 1000, disabling/enabling
> SMT scheduling, dual core scheduling, using deadline scheduler, using
> Ingo Molnar's realtime kernel patch, disabling APIC, but non of them
> helped.

It's more likely that alsa is misbehaving than anything else.
Try the oss emulation of alsa, it's a lot more reliable than
using alsa directly. But as i said above, it could be virutally
anything, so please provide us with more information.
(and move the discussion over to -users)


			Attila Kinali
 


-- 
Why does it take years to find the answers to
the questions one should have asked long ago?



More information about the MPlayer-dev-eng mailing list