[MPlayer-users] Re: Radeon frambuffer use if Radeon is not primary displThu Dec 13 01:29:45 2001
Arpi
arpi at thot.banki.hu
Thu Dec 13 00:34:02 CET 2001
Hi,
> > Btw later they found that AGP is slow too, and instead of placing textures
> > in system ram (and access via AGP) they started to increase amount of memory
> > on the card -> 64/128mb radeons/geforces.
>
> Which of course means that for an Mplayer/video-out only display
> device, the minimum available memory that can be bought (I have not
> seen less than 32MB cards these days) is all that is needed. I would
> bet that most of it would be wasted for video-out.
>
for video playback, you need max. 2*xres*yres*3 bytes and some megs for the
desktop (at 32bpp it's 4*xres*yres) and XAA (128kb ?).
so, in general, a 8mb card should be enough for everything.
i had mem size problems with 4mb ati in my laptop with 720x576 res dvds
at 1024x768x32bpp.
> > but todays cards like matrox g series and radeons are all AGP chips. the PCI
> > versions of these cards (if exists)
>
> They certainly do exist.
at leats matrox does produce pci versions, i don't know about radeon.
> > have PCI-to-AGP bridge on teh card and
> > the AGP chip... and they aren't supported under linux afaik.
>
> No? The radeon_vid frame-buffer driver that you have in the Mplayer
> project would not support a PCI Radeon card?
i think no.
there was a discussion about such cards at DRI-devel list some months ago.
the driver should know and control the pci-to-agp bridge on the card to
reach the video chip. and current drivers can't do that...
> So being cost concious, what is the most "featureless" Radeon that
> would be acceptable for "Mplayer"ing? I have an ATI TV-Wonder tuner
> card right now, but that Radeon with the TV-In and the TV-Out is
> looking mighty nice. :-)
why do you need radeon? what about an oldy mach64 chuip card with tvout?
it's even better supported by gatos drivers.
A'rpi / Astral & ESP-team
--
mailto:arpi at thot.banki.hu
http://esp-team.scene.hu
More information about the MPlayer-users
mailing list