AMD Releases Cayman Documentation, Open Driver Is Close

[Drago]

Yes, Liano is a nice piece of silicon(at least on paper), and I am waiting it for my next build.
AMD are talking about Cayman for the next merge window, and not a word about Liano. That is why I hope Bridgman shows up, to clarify the situation.

S3TC is implemented by external library onli for r300g, but not for r600g.

[89c51]

What I'm trying to say is that the patent will stop the drivers from having these features, because no company or well-known open source organization is going to work on implementing patented software.

not all graphic hackers are paid by RH, xorg or other organizations. I think Corbin and Marek + some nouveau guys (and others) are not on they payrol of any company afaik and they have contributed a lot of code. These people have the knowledge to write driver stuff and probably can write the code needed.

i also seem to remember that S3TC has been implemented (fully or partially i don't know) in an external library or something

[Adarion]

Wasn't it LLano (instead of L i ano)?

[Drago]

uhh you don't get an important point.

r800 vs bobcat is a difference they need to dev some stuff.

but there is no difference between bobcat and Liano on the driver side.

and yes adding an PCI ID is really a hard work

You are really smart guy, aren't you?
Yes, pci ids are piece of cake, but they will come to 2.6.40. 3 months after the HW starts shipping. I really hope AMD will do some Liano commits along with Caymans.

[Drago]

You smart guy... Bobcat is 5xxx GPU also, and it needed AMDs attention to get it working, so will be Liano. The pci ids will be added at least, but that work will be headed towards 2.6.40 in the september maybe. So no out of the box experience for Liano.

[libv]

Nice to see that AMDs streamcomputing/GPGPU people are still keeping the documentation dream alive.

Thanks guys!

[blackshard]

Well AMD, I am pleased for the speed for this new chunk of documentation. I can say I am all yours
One thing what bothers me, is why do you prefer to implement Caymans first, instead of Liano. No one with Cayman, as a highest end GPU from AMD, is gonna use r600g driver. No one would pay >$300, and use it as a GPU for <$100. Now this work will probably land in 2.6.39, which would be the kernel when Liano start showing on the market. I am headed towards Liano, after a couple of months, and this is somewhat disappointing. Bridgman, feedback appreciated.

I think Llano will use hd5xxx. It would be easy to adapt current r600 driver. Llano will also ship in Q3 2011 AFAIK, so there's time to work on.

[Drago]

It's not the former topic, but about S3TC : what is the needed work to have libtxc-dxtn handled by r600 ? Because, it's sad, I know r600g can run commercial game (eg. ETQW which I have tested) but is totally useless without S3TC... I'm ready to pay for that !

Michael, could you do an investigation on S3TC and r600 support ?

+1
With everything now working, S3TC is becoming pin in the *ss.
May be there should be campain with donations, exactly for this feature. I am willing to pay also.

[whitecat]

S3TC and r600

It's not the former topic, but about S3TC : what is the needed work to have libtxc-dxtn handled by r600 ? Because, it's sad, I know r600g can run commercial game (eg. ETQW which I have tested) but is totally useless without S3TC... I'm ready to pay for that !

Michael, could you do an investigation on S3TC and r600 support ?

[Drago]

Well AMD, I am pleased for the speed for this new chunk of documentation. I can say I am all yours
One thing what bothers me, is why do you prefer to implement Caymans first, instead of Liano. No one with Cayman, as a highest end GPU from AMD, is gonna use r600g driver. No one would pay >$300, and use it as a GPU for <$100. Now this work will probably land in 2.6.39, which would be the kernel when Liano start showing on the market. I am headed towards Liano, after a couple of months, and this is somewhat disappointing. Bridgman, feedback appreciated.