On Sun, Aug 04, 2002 at 12:37:51PM +0200, Denis Oliver Kropp wrote:
> This sounds really cool, but I don't know if it works with my G550, too.

I only have access to a G400 DH so someone will have to test it.

> One question: Does it override the secondary framebuffer device?

Yes. All crtc2 programming is completely done in DirectFB. Maven is
also programmed from DirectFB but via /dev/i2c. Only thing that would
require additional kernel stuff is vsync (or more precisely vblank in 
this case). I was thinking about passing a flag to the crtc1
FBIO_WAITFORVSYNC. What do you think?

> Having single
> head support in matroxfb only with full access to both heads and
> memory in DirectFB seems to be an acceptable solution. We just have to
> add support for layers on another screen (output)...

That's what I figured.

One problem is that it won't work with a monitor. So if someone wants to
use on monitor w/ crtc2 they'll have to use matroxfb_crtc2. And that won't
allow multihead operation.

One other thing is that 656 mode forces interlaced operation. I'm not sure
that's a problem though. According to the specs it should support
non-interlaced operation in all but ARGB mode but I was unable to make it
work. I believe this is simply because maven assumes interlaced data.

The biggest drawback might be the fixed resolution. All applications that
want hardware scaling will have to use a temporary surface and do a
stretchblit from there. That's what I assume is done even in Windows since
crtc2 was using ARGB mode while I was playing some divx videos. I was
thinking about the possibility of making this transparent to the user.
That would mean having two surfaces allocated and doing a stretchblit
inside the driver. Any ideas?

-- 
Ville Syrj�l�
[EMAIL PROTECTED]
http://www.sci.fi/~syrjala/


-- 
Info:  To unsubscribe send a mail to [EMAIL PROTECTED] with 
"unsubscribe directfb-dev" as subject.

Reply via email to