[vdr] Re: Softdevice + EPIA-M / Sound resampling

["Colin Paton" <ac2crp@blueyonder.co.uk>]

Hi,

> With most framebuffer devices you could set the videomode (not vesafb)
> through ioctls so with viafb I think this could be done. Maybe there
> is a wrapper in DirectFB to do this too.
> BTW, what is the vertical frequency of the mode you're using when
> testing? If it's not matched as close as possible to the frame rate in
> the stream
> you would defenately run in to trouble with the VBL  syncing.

As far as I'm aware (from comments on the DirectFB mailing list) the clocks
are actually generated by the vt1622 TV encoder chip. The CRT controller is
programmed to be slaved to an external clock from the vt1622.

It does appear to be generating 25 frames per second however, which is what
I want. The code currently does not seem to be synchronised to the correct
field, so you timetimes get an interlace blur effect. As a workaround for
now I'll tell the video chip to deinterlace the image, however it would be
better to set the TV controller correctly.

I've managed to record a few samples of a channel which has a scrolling
message at the bottom of the screen - this will be useful for testing as
freezes/dropped frames are immediately apparent.

Does anybody know if the new IO schedulers in Kernel 2.6 are likely to
improve things? I'm sure that there is occasionally a dropped frame still.
I'm also wondering if setting the thread scheduling priority would improve
things - at the moment all threads have equal priority and use the default
scheduler rather than any realtime scheduling.

Thanks,

Colin



-- 
Info:
To unsubscribe send a mail to ecartis@linuxtv.org with "unsubscribe vdr" as subject.