[mythtv] still getting jitter/speed pulsing

[Mark Spieth]

Interesting observations. I have also been chasing this.
I use busywait sync as none of the others work.
what I have found is the call to show (XSync) is effectively a flush to X of
the currently written page.
I assume this is aligned with sync as it happens at that rate.
my output display (monitor) is at 60Hz.
my input frame rate is at 25Hz (pal)
I have found that the minimum period between XSyncs is 33ms (2 output
frames)
and increases at multiples of the frame period (16.66ms)
if the desired frame rate is close or > than the output framerate/2, then
strange things start to happen.
try using timestretch so that input rate > outputrate/2 and you see it
clearly.
for me this is > 1.2 as 25*1.25 > 60/2, 1.2 seems to work ok.
frames must be deliberately skipped to compensate for this.
I still dont have a good solution yet.
havent updated to latest ffmpeg sync yet so still working off 24/12
reverting libavcodec and libavformat back to 29/11 does nothing.
still very confused as I dont fully understand the underlying xv behaviour.
cheers
mark

> The difference appears to be that the jitter happens only in the
> presence of both Xv output and the new ffmpeg code.  I don't know how
> the two can possibly be related, which is part of the reason I'm
> having a hard time figuring out where to start debugging this.
>
> I added a few Jitterometers in AVSync, and found a potentially
> interesting difference between X11 and Xv output.
>
> X11 (normal)
> ------------
> 'prepare1' sd0fmean = '12.87', mean = '8.21', std. dev. = '1.06', fps =
'121802.68'
> 'vsync1' sd0fmean = '11.00', mean = '4.67', std. dev. = '0.51', fps =
'214132.76'
> 'vout1' sd0fmean = '17.59', mean = '14254.70', std. dev. = '2506.94', fps
= '70.15'
> 'prepare2' sd0fmean = '10.90', mean = '7.83', std. dev. = '0.85', fps =
'127713.92'
> 'vsync2' sd0fmean = '12.76', mean = '3.81', std. dev. = '0.49', fps =
'262467.19'
> 'vout2' sd0fmean = '13.14', mean = '15888.00', std. dev. = '2088.11', fps
= '62.94'
> 'video_output' sd0fmean = '2.15', mean = '33318.28', std. dev. = '716.07',
fps = '30.01'
>
> Xv (abnormal)
> -------------
> 'prepare1' sd0fmean = '9.16', mean = '7.82', std. dev. = '0.72', fps =
'127877.24'
> 'vsync1' sd0fmean = '103.59', mean = '12139.34', std. dev. = '12575.42',
fps = '82.38'
> 'vout1' sd0fmean = '9.75', mean = '3117.61', std. dev. = '304.00', fps =
'320.76'
> 'prepare2' sd0fmean = '12.86', mean = '7.06', std. dev. = '0.91', fps =
'141643.06'
> 'vsync2' sd0fmean = '25.75', mean = '12002.08', std. dev. = '3090.27', fps
= '83.32'
> 'vout2' sd0fmean = '10.26', mean = '2886.44', std. dev. = '296.08', fps =
'346.45'
> 'video_output' sd0fmean = '43.54', mean = '33651.20', std. dev. =
'14652.30', fps = '29.72'
>
> Note that the first vsync seems to vary a *lot*.  I would expect vsync
> in Xv to take more time, since the frames actually display quicker so
> mythfrontend is spending more time idle; but the standard deviation is
> as high as the mean, which suggests something funky is going on there.
>
> What I don't understand is why vsync would start acting up just
> because I started using Xv output: I'm using GL sync in both cases.
>
> Any thoughts?
>
> Cheers,
> Kyle
>


----------------------------------------------------------------------------
----


_______________________________________________
mythtv-dev mailing list
mythtv-dev at mythtv.org
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev



-- 
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.298 / Virus Database: 265.6.7 - Release Date: 12/30/2004