News ATI drivers

Luc Verhaegen libv at skynet.be
Tue Sep 11 15:49:49 PDT 2007


On Mon, Sep 10, 2007 at 10:46:11AM -0700, Adam Williamson wrote:
> On Mon, 2007-09-10 at 10:39 -0700, Alan Coopersmith wrote:
> > Adam Williamson wrote:
> > > Anyway, to get back on topic - now it's Monday, any news from any
> > > ATI-involved X developers? It would be nice to know what's going to
> > > happen with avivo and so forth.
> > 
> > http://airlied.livejournal.com/50187.html
> > http://www.fooishbar.org/blog/tech/x/amd-2007-09-10-09-22.html
> > http://www.x.org/wiki/Events/XDS2007/Notes
> 
> I rather meant opinions and plans from the people who were *already*
> working on ATI drivers, but thanks. :)

Well,

We spent most of the weekend either travelling or coding. On monday we 
spent most of the time coding at XDS, and leaving very early at the 
sponsored supper. Today we spent most of the day coding in our 
hotelroom, and only made it very late to XDS (in the middle of daniels 
his talk - sorry daniel, takes 25-30 minutes on foot to get there from 
the hotel).

But on the plus side, this m56 laptop now has panel support (unscaled, 
and i cut some corners in the restore routine), and DDC and everything 
now is all somewhat tied together. Enough to make the driver throw up 
its hands in the air when it validates the modes of the panel and the 
VGA connector to the projector at XDS (we of course don't have access to 
other monitors while out here) :)

Panel = single fixed mode at 1680x1050
Projector = up to 1280x1024

There is just no way that they will support the same mode at the same 
time :)

So at least it is no longer doing wrong things, (since around 18.00 
today), but i still need to stop it from giving up in this case.

Egbert is currently working on retrieving connector mapping information 
out of the atombios tables. I'm currently hardcoding them in a pci 
subsystem id based table, and doing this for every card out there is, 
hrm, kind of timeconsuming :)

Anyway, we know that this is taking long, but we want to be somewhere 
when we push out our code as we will be swamped in userreports for a few 
weeks as soon as we do send this out. Given the nature of the hardware 
(you won't believe all the fiddly bits that need to be set up), and 
the fact that we currently know 96 pci-ids (not counting the r6xx 
firegls yet) there is a lot of pain to come.

It certainly is not going to be finished or shiny or whatever, but we 
don't want to put us, or anyone else (like the users testing this new 
code) through more trouble than needs be.

Luc Verhaegen.



More information about the xorg mailing list