ATI R800 linux

[Panix]

Why is everybody so obsessed with OpenGL speeds, when even simple 2D stuff is unbearably slow. I can't work normally in Linux. I have a 5870, with 10.7 installed. If I enable the new 2D code, I have black or gray planes allover (mostly when using firefox or thunderbird). And switching desktops is really slow, as it needs to redraw the whole screen. (I simply have metacity as wm, nothing fancy).

Is there a roadmap of the driver development somewhere? I'd like to know if this is hopeless, or if this will be fixes withing the next 2 or 3 driver releases.

Join the club. It seems the HD 5xxx owners who "complain" are ignored but the owners who are ambiguous about what works receive replies from the AMD guys... hmmmmmm....

Go to the Ubuntu forums. Ubuntu is supposedly one of the distros ATI supports with the binary driver but there are tons of threads/posts about the black screens or resolution being very low. Apparently, FOSS driver still don't work (right?) with these cards. At least, not 3D. It sounds like 2D is slow. But, there's no ETA given... for anything?

[V!NCENT]

Shadowfb is murdering 2D on my HD5770. What are you talking about?

[bridgman]

Join the club. It seems the HD 5xxx owners who "complain" are ignored but the owners who are ambiguous about what works receive replies from the AMD guys... hmmmmmm....

In case you are honestly confused and not just trolling, people who ask about features which *have* been released get answers, while people who ask about features which have not been released do not get answers other than the occasional reminder that we do not talk about unreleased software features.

[Panix]

In case you are honestly confused and not just trolling, people who ask about features which *have* been released get answers, while people who ask about features which have not been released do not get answers other than the occasional reminder that we do not talk about unreleased software features.

Okay, my comment was a bit harsh. I'm not trolling or don't think I am.

Is there a way to summarize the current state of the R800 cards as far as the binary driver goes?

2D - ???? have to use OpenGL output? etc. etc.? no video decoding/no hardware acceleration... xv doesn't work... ??

3D - ???

Or is there a situation of development in that there's a bunch of features being worked on by developers?

Also, is there the kernel and xorg version restrictions and will this continue (as long as it takes time to develop the drivers and technologies)?

Anyway, carry on for the specific on-topic discussion.

[PGHammer]

BInary drivers and AMD R8xx

Okay, my comment was a bit harsh. I'm not trolling or don't think I am.

Is there a way to summarize the current state of the R800 cards as far as the binary driver goes?

2D - ???? have to use OpenGL output? etc. etc.? no video decoding/no hardware acceleration... xv doesn't work... ??

3D - ???

Or is there a situation of development in that there's a bunch of features being worked on by developers?

Also, is there the kernel and xorg version restrictions and will this continue (as long as it takes time to develop the drivers and technologies)?

Anyway, carry on for the specific on-topic discussion.

If you are running either 'buntu or openSuSE and the released (not development) version (Lucid Lynx in the case of 'buntu, or 11.3 in the case of openSuSE), you'll want AMD Catalyst 10.7 or 10.8 (both versions support the current kernels in these two distributions). Both will also support the development versions (Maverick Meerkat and openSuSE 11.4 milestone 1) *if* you have not updated to Xorg 1.9 and/or Mesa 7.8.x. (Xorg 1.9 ships as an update for Maverick Meerkat beta 1, and is included with openSuSE 11.4 milestone 1, hence my not recommending either if 3D is a necessary.)

If your needs can be met *without* 3D, Maverick and Xorg 1.9 has a great deal to recommend, even though 3D doesn't work with anything newer than R5xx reliably right now. First off, the combo of the radeon driver in development (xorg-edgers PPA and x-swat PPA) and 1.9 have blistering fast 2D (actually faster than Catalyst 10.7 or 10.8 2D in Lucid), and I have not heard anyone explain why (or how) that is (the how is pretty darn important); for that reason, since I have no use for 3D right now, I'd actually rather run Maverick than Lucid, and for performance reasons.

[Panix]

If your needs can be met *without* 3D, Maverick and Xorg 1.9 has a great deal to recommend, even though 3D doesn't work with anything newer than R5xx reliably right now. First off, the combo of the radeon driver in development (xorg-edgers PPA and x-swat PPA) and 1.9 have blistering fast 2D (actually faster than Catalyst 10.7 or 10.8 2D in Lucid), and I have not heard anyone explain why (or how) that is (the how is pretty darn important); for that reason, since I have no use for 3D right now, I'd actually rather run Maverick than Lucid, and for performance reasons.

Interesting. That brings me to a question I keep asking repeatedly but never getting an answer. Bridgman refuses to answer.

The Evergreen cards were released last Sept/Oct '90 so almost a year ago. Windows drivers work, both 2D and 3D. The comparison here is 3D. If the Catalyst 3D part of the driver doesn't work reliably, then the question is why. It's understandable and acceptable if the open source driver's 3D capabilities are still in flux but no excuse for the binary driver to still be a problem. Aren't they just borrowing some code from the Windows stack driver? Nvidia does this, right? Maybe ATI does something different but still, after a year?!? It shows the resources aren't there? AMD/ATI don't make it a priority. There's little resources for the open source but maybe they want that way for both drivers?

The religious zealots of the ATI solution preach how great ATI is but this is constantly shown to be the opposite. The support just...isn't...there.

[PGHammer]

The AMD Driver Issue (in Linux)

Interesting. That brings me to a question I keep asking repeatedly but never getting an answer. Bridgman refuses to answer.

The Evergreen cards were released last Sept/Oct '90 so almost a year ago. Windows drivers work, both 2D and 3D. The comparison here is 3D. If the Catalyst 3D part of the driver doesn't work reliably, then the question is why. It's understandable and acceptable if the open source driver's 3D capabilities are still in flux but no excuse for the binary driver to still be a problem. Aren't they just borrowing some code from the Windows stack driver? Nvidia does this, right? Maybe ATI does something different but still, after a year?!? It shows the resources aren't there? AMD/ATI don't make it a priority. There's little resources for the open source but maybe they want that way for both drivers?

The religious zealots of the ATI solution preach how great ATI is but this is constantly shown to be the opposite. The support just...isn't...there.

Part of the problem with the various Linux distributions is that each one has its own *quirks* (while sometimes, those quirks can be adjusted to with scripting changes, sometimes, it requires a complete development cycle to deal with). In fact, let's look at the changes in Ubuntu alone over 10.04 and 10.10. As I stated before, 10.04 can use the binary 3D driver (even for Evergreen); in fact, 10.7 and 10.8 are the first two releases of the Linux binary driver to support Evergreen for 3D. (I have HD5450, AKA Cedar, Evergreen's budget part, so this is not trivial to me in the least, and it's also something I personally am aware of.) Even 10.10 can use Catalyst 10.7 and 10.8 (though 10.8 is preferred) as long as you don't upgrade to Xorg 7.9 (which appears as an update for Maverick beta 1). There are changes (and pretty darn serious ones) from Xorg 1.8 to Xorg 1.9 (and they cause issues with software other than AMD's binary driver; these same changes seriously bork KDE 4.5.1 as well, and that is with the vanilla radeon driver, without 3D acceleration). Throw in that Ubuntu is also using a new Mesa (Ubuntu will likely be the first major distribution to deploy Mesa 7.9) and AMD really is trying to hit a moving target. (While nVidia has drivers that work *now*, there is no guarantee that they will continue to work; even the ones that work are no more than beta drivers.) The changes between 10.04 and 10.10 are far greater than between 8.04 and 9.04, for example.

[Panix]

Part of the problem with the various Linux distributions is that each one has its own *quirks* (while sometimes, those quirks can be adjusted to with scripting changes, sometimes, it requires a complete development cycle to deal with). In fact, let's look at the changes in Ubuntu alone over 10.04 and 10.10. As I stated before, 10.04 can use the binary 3D driver (even for Evergreen); in fact, 10.7 and 10.8 are the first two releases of the Linux binary driver to support Evergreen for 3D. (I have HD5450, AKA Cedar, Evergreen's budget part, so this is not trivial to me in the least, and it's also something I personally am aware of.) Even 10.10 can use Catalyst 10.7 and 10.8 (though 10.8 is preferred) as long as you don't upgrade to Xorg 1.9 (which appears as an update for Maverick beta 1). There are changes (and pretty darn serious ones) from Xorg 1.8 to Xorg 1.9 (and they cause issues with software other than AMD's binary driver; these same changes seriously bork KDE 4.5.1 as well, and that is with the vanilla radeon driver, without 3D acceleration). Throw in that Ubuntu is also using a new Mesa (Ubuntu will likely be the first major distribution to deploy Mesa 7.9) and AMD really is trying to hit a moving target. (While nVidia has drivers that work *now*, there is no guarantee that they will continue to work; even the ones that work are no more than beta drivers.) The changes between 10.04 and 10.10 are far greater than between 8.04 and 9.04, for example.

Okay, I think I understand. That makes sense. I also acknowledge some issues reported in the nvidia nvnews forums regarding use of Xorg 1.9 and the new Fermi-based cards. So, yes, I believe there's uncertainty there and their method is definitely not perfect notwithstanding the proprietary driver limitation (more or less).

I'm just wondering why such delay in being able to adapt and adjust to such changes in Xorg or whatever effects the driver. I suppose it's a bunch of intangibles, the difference in the binary driver(?) compared to nvidia(?) and not as many resources in proportion (my assumption). Anyway, thanks for explanation. I meant to be more inquisitive than confrontational (if that's what it sounds like). I want to believe an ATI solution is a good investment as I've mentioned before.

[Panix]

I read these types of reports quite often:

http://www.phoronix.com/forums/showthread.php?t=25987

http://www.phoronix.com/forums/showthread.php?p=146800#post146800

But, I give thanks to the replies of my questions. I think I'm just going to wait now and save up for a card. I think there's plenty of info and references here so I can decide when the time comes. I understand many here are tired of my questions so I'll just read the posts for a bit. Thanks for the info! Cheers!

[bridgman]

Interesting. That brings me to a question I keep asking repeatedly but never getting an answer. Bridgman refuses to answer.

Panix, I answer the first time, sometimes the second, but by the third time you ask the same question I usually get tired of answering. You also ask nearly as many questions as everyone else on the forum combined.

The Evergreen cards were released last Sept/Oct '90 so almost a year ago. Windows drivers work, both 2D and 3D. The comparison here is 3D. If the Catalyst 3D part of the driver doesn't work reliably, then the question is why. It's understandable and acceptable if the open source driver's 3D capabilities are still in flux but no excuse for the binary driver to still be a problem. Aren't they just borrowing some code from the Windows stack driver? Nvidia does this, right? Maybe ATI does something different but still, after a year?!? It shows the resources aren't there? AMD/ATI don't make it a priority. There's little resources for the open source but maybe they want that way for both drivers?

The religious zealots of the ATI solution preach how great ATI is but this is constantly shown to be the opposite. The support just...isn't...there.

In this case you seem to be hypothesizing substantially different OpenGL behaviour between the Windows and Linux drivers, then interpreting that as a "lack of support for Linux", is that correct ?