Well, There's No Radeon HD 6000 Open-Source Surprise

[phoronix]

Phoronix: Well, There's No Radeon HD 6000 Open-Source Surprise

Earlier this week AMD launched the Radeon HD 6850 and Radeon HD 6870 graphics cards as their first next-generation offerings that belong to the "Northern Islands" family. A day later there was the Catalyst 10.10 Linux driver release from AMD that added support for these new "Barts" GPUs to their proprietary driver, but no open-source support has yet to be found...

http://www.phoronix.com/vr.php?view=ODcxMg

[libv]

Wasn't atombios going to make basic hw bringup so trivial that new hardware would be irrelevant and that it would work almost out of the box?

Clued people have always known that this was never going to be the case, and apparently, time is catching up with Bridgie here...

Back in the radeonhd days, we got bashed to no end because we "wasted time reimplementing atombios"... The fact of the matter was, we spent 10-20% of our time on REing atombios (which we needed to do anyway, since we never ever got told how to use it), and the rest we spent figuring out nasty issues to which Bridgman was never able to give us any solid answer.

By the time the radeon competition came up, they spent their time ignoring new hw bring-up and played with things like r500 3d support, and patiently waited until the major bumps were solved by the radeonhd guys. Of course then not having to write the "standard" register banging code sped up things: there was nothing left to do but pick up the hard work of the radeonhd guys...

So... Where are we today. There are no more guys like me doing the real work and still getting the blame. So, now excuses have to be made up, or the truth might come out...

Maybe atombios is not the wonderous solution that it was always claimed to be. Maybe it really is an extra, untouchable, layer in between, which can be as much of a part of a solution as it can be the problem.

[Melcar]

Bummer. Though to be honest, I never expected speedy open source support for new chips. Never really cared about AtomBIOS either, except that I didn't really "get it" in terms of usefulness and always liked the way radeonhd was doing things.

[chithanh]

I wonder whether the "non-trivial changes in the display portion" are related to DisplayPort 1.2 multiple streams.

[bridgman]

The changes are in more fundamental areas than that as well, things like the lookup tables. We'll know more once Alex has a chance to poke at the display pipe a bit.

Our early Northern Islands (NI aka HD6xxx) work was mostly trying to get a head start on the bigger 3D changes and make sure that jumping to the Gallium3D code base was going to work OK... unfortunately the first NI part launched earlier than I was expecting and the first Fusion part launched later.

[TemplarGR]

Wasn't atombios going to make basic hw bringup so trivial that new hardware would be irrelevant and that it would work almost out of the box?

Clued people have always known that this was never going to be the case, and apparently, time is catching up with Bridgie here...

Back in the radeonhd days, we got bashed to no end because we "wasted time reimplementing atombios"... The fact of the matter was, we spent 10-20% of our time on REing atombios (which we needed to do anyway, since we never ever got told how to use it), and the rest we spent figuring out nasty issues to which Bridgman was never able to give us any solid answer.

By the time the radeon competition came up, they spent their time ignoring new hw bring-up and played with things like r500 3d support, and patiently waited until the major bumps were solved by the radeonhd guys. Of course then not having to write the "standard" register banging code sped up things: there was nothing left to do but pick up the hard work of the radeonhd guys...

So... Where are we today. There are no more guys like me doing the real work and still getting the blame. So, now excuses have to be made up, or the truth might come out...

Maybe atombios is not the wonderous solution that it was always claimed to be. Maybe it really is an extra, untouchable, layer in between, which can be as much of a part of a solution as it can be the problem.

Why not cooperating? Do you place your ego above the good of the opensource community?

If there are good guys like you who know how to write for the radeon drivers, and are experienced in it, why they are sitting idle? Because it is not RADEONHD and it is RADEON? WTF?

Believe me, i am using the radeon driver for almost a year, on an ΑΤΙ ΗD3870, and if i could find the time needed i would try to contribute myself, even if this meant learning quite a bit of information before i am able to produce one line of code for it...

If you old RADEONHD devs can do something, then please, do. These drivers are vital for our community.

[dfx.]

damn, even they admitting that only thing keeping DRM alive is 'security'-by-obscurity principle, secrets and collusions. and they still doing it...

we could use 100% of hardware in devices we 100% payed for but instead we forced to obey interests of people we don't like on platforms we don't use for purposes we despise.

[libv]

Why not cooperating? Do you place your ego above the good of the opensource community?

WTF?

So, open source mindset (like, free and open hw documentation), and solid maintainable code (hey, like the radeonhd driver), versus egos (like, "i can write the modesetting support in 1500loc", and the need to do bling work so that big announcement can be made).

Just who are you really bashing here, or what part of history are you denying?

Next up, Templar claims there were no concentration camps in WWII!

If there are good guys like you who know how to write for the radeon drivers, and are experienced in it, why they are sitting idle? Because it is not RADEONHD and it is RADEON? WTF?

I am not sitting idle, far from it. And maybe here too you should try to gather some historical facts.

Believe me, i am using the radeon driver for almost a year, on an ΑΤΙ ΗD3870, and if i could find the time needed i would try to contribute myself, even if this meant learning quite a bit of information before i am able to produce one line of code for it...

Yeah, great, whatever...

If you old RADEONHD devs can do something, then please, do. These drivers are vital for our community.

Where were you 2-3 years ago?

[Tgui]

I read this block of text as:

AMD's John Bridgman, who heads the their open-source work, is a member of our forums (along with AMD's Alex Deucher and many other developers). He's not only a member, but a very attractive member, and in fact the most attractive member in our entire community with more than 5,200 posts

I was wondering if Michael kissing ass would get better results out of John.


Sorry for the off topic here!

On topic, its sad to see the state of ATI open source drivers. I certainly don't fault the devs, they do what they can with what they have given how daily life can mess up grand plans. I really wish AMD/ATI would throw more money at this situation and get more people on their open source drivers. Hell, more stable binaries would be a good start as well!

I'm rooting for you AMD! I've been burned a few times by your drivers in the past and now use Intel exclusively for my graphics and NVIDIA for OpenCL/CUDA. Not that it counts for much, but I would gladly give you my money if you had a truly open source feature complete 3D driver!

[pingufunkybeat]

Out of interest: what is the Intel driver providing for you that the ATI driver does not provide?