On Mon, Dec 18, 2000 at 03:55:36PM +0100, Andreas Beck wrote:
> > A question about the tile display driver. Has anybody tried a quick and
> > durty support of the VESA v1.2 standard with only windowed memory (no linear
> > frame buffer) with the tile visual on top of it?
>
> Umm - what would you want to achieve with that ?
>
> Kind of "caching" the changes to avoid having to pageflip often ?
Yes and could draw an entire part of the screen before an other instead of poking
here and there the memory depending of the drawing function.
>
>
> > Does the tile display driver perform any kind of optimizations before
> > accessing underlying display drivers?
>
> Yeah - at least when directbuffers are in use, the tile target allocates
> Main memory and blits to the real target in regular intervals.
>
> This would - in that specific case - count as an optimization.
Exact. So let's give it a try!
>
> How does *BSD handle such paged buffers ? Like we did in scrdrv 0.7 by
> making a virtual linear buffer from it using the MMU ?
There's nothing special as far I know. Just an ioctl to change the windows
(A or B) origins. Really, you did this wonderfull thing of the MMU? Cool!
What's the performance with that? Horrible or acceptable?
So, as a demonstration of tile power and to check FreeBSD support I'll do this
with tile. Later, for the KGI port, I'll consider the MMU solution.
Nicholas
--
[EMAIL PROTECTED]
Alc�ve - Open Source Software Engineer - http://www.alcove.fr
