[MPlayer-dev-eng] [PATCH] Improved DGA vertical retrace sync

[Arpi]

Hi,

> Not in the current implementation, except that Mplayer seems to drift
> between the number of draw_frame calls compared to the number of
> flip_page calls especially when fast-forwarding, so four or five

could you detail this? i don't understand :(

> If Mplayer sometime would support a more advanced video driver API,
> more than three frames could make much sense. Consider the case where
> Mplayer pre-renders 12 frames ahead, and a sudden temporary disruptive
> scheduling problem occurs (like heavy disk activity or whatever).

see mplayerxp or xine (yes i know you're already work on xien too)

we do not plan multithreading (separated decoding and display) in mplayer.

> > - it should honor the -double/-nodouble flag, and by default (nodouble)
> > use only one buffer with no buffer flipping. you've just removed the
> > if(!vo_doublebuffering) check.
> 
> Sorry. The attached patch below should fix this. Why have you selected
> nodouble as default? It does not make much sense to me.

for DGA, you may be right. but i remeind you that soem dga driver doesn't
support well teh page flipping.
and, for xv or some other, double buffering needs extra confitions or may
cause problems. let teh user to enable it.

> > most vga cards can trigger an IRQ at vsync, it should be possible to 
> > develop a kernel driver which can catch that. 
> 
> Well, yes... But (assuming we still talk about the DGA interface), this
> would have to be implemented in the XFree86 server and since they plan
> on removing the DGA extension in XFree86 5, they will probably not go
> ahead with something like that.

it can't be done well in user space - requires kernel helper to handle the
interrupts. xfree already has it: drm.


A'rpi / Astral & ESP-team

--
Developer of MPlayer, the Movie Player for Linux - http://www.MPlayerHQ.hu