Intel: how's dynamic framebuffer allocation coming?

[Joel Feiner]

Just to satisfy my curiosity, why was it that with MergedFB on radeon
(before randr 1.2) I was able to have full acceleration across two screens?
Was there no scan out limit in that particular case or was there some other
workaround?

On Thu, Oct 16, 2008 at 1:39 PM, Adam Jackson <ajax at nwnk.net> wrote:

> On Tue, 2008-10-14 at 21:28 +0100, Andrew Clayton wrote:
> > On Mon, 13 Oct 2008 17:08:29 -0400, Joel Feiner wrote:
> >
> > > And on that note, how about allowing proper 3d and compositing
> > > acceleration across screen area bigger than 2048x2048 (I believe
> > > that's the limit).  It'd be nice to be able to use Compiz/xcompmgr
> > > with two screens again like I could with MergedFB.
> >
> > Adam Jackson has a project called shatter which should get around
> > this. See last paragraph here:
> > http://ajaxxx.livejournal.com/60080.html
>
> There's a (broken, incomplete, useful for reading but not for running,
> please don't even try to run it) version of this idea here:
>
> http://cgit.freedesktop.org/~ajax/xserver-shatter?h=shatter<http://cgit.freedesktop.org/%7Eajax/xserver-shatter?h=shatter>
>
> In case anyone's interested in the general shape of the idea.  The
> comment at the top here is, well, more or less accurate:
>
>
> http://cgit.freedesktop.org/~ajax/xserver-shatter/tree/randr/rrmx.c?h=shatter&id=2d97ccc7ef20ab186dbe681fbe101ab4a246edfd<http://cgit.freedesktop.org/%7Eajax/xserver-shatter/tree/randr/rrmx.c?h=shatter&id=2d97ccc7ef20ab186dbe681fbe101ab4a246edfd>
>
> The big work left to do is splitting this the rest of the way out from
> RANDR, redoing rendering dispatch so that everything is against
> contexts, probably some other stuff I'm forgetting.
>
> I've got a better writeup on the core API changes this will need, and a
> less garbage implementation, laying around on some laptop somewhere.
> I'll see about getting those uploaded.
>
> - ajax
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.x.org/archives/xorg/attachments/20081016/fb072023/attachment.html>