AMD Linux Catalyst: Hardware Owners Screwed?

[bridgman]

Regarding hardware support, the nouveau driver had 0-day support for the Kepler-GPUs. What about radeon's support for the HD7000 series when they were released.

A better question would be radeon's support for Trinity (which had a comparable change in programming model to Kepler), where support was released a couple of months before hardware became available.

[Kano]

Did you ever compare trinity against ivy bridge? Which chip has better oss day-0 support? Dont forget vaapi...

[smitty3268]

Did you ever compare trinity against ivy bridge? Which chip has better oss day-0 support? Dont forget vaapi...

Trinity was probably more comparable to Sandy Bridge launch. And that's not so bad, considering it's the first time they've managed to catch up to new hardware.

Obviously Ivy is ahead, but the point is that AMD is still trying to catch up.

[hoohoo]

i personally use Kubuntu right now and yes unity sucks balls.

I'm unhappy with "Fedora" because I'm not the slave for red-hat's Enterprise "customers"
Fedora isn't a Linux distribution at all its just a way to abuse free-software in a business model.

maybe i switch to Debian 7 wen it comes out. because KDE is in death hands on the ubuntu side.

"using the 3870.... the card is semi dead Right Bloody Now."

do you mean the catalyst or do you mean the radeon driver? the catalyst was death on Fedora at all times the radeon driver should work.

Getting pretty off topic, but FWIW OpenSuSE 11 and 12 have very good add-on repo offering KDE 3.5 desktop. The repos are semi-official and the packages are quite stable.

Something tells me you're not a KDE 4.x kind of guy...

[Qaridarium]

Getting pretty off topic, but FWIW OpenSuSE 11 and 12 have very good add-on repo offering KDE 3.5 desktop. The repos are semi-official and the packages are quite stable.

Something tells me you're not a KDE 4.x kind of guy...

i like kde 3,5 very much but i use kde4.x without pain

[Dandel]

http://www2.ati.com/drivers/hotfix/c...x86.x86_64.zip

12.6

edit.

Instaling them right now, lets see if they work with my HD4850.

edit2.

Well, they do not work with my HD4850. Buing nvidia next time.

I would actually suggest avoiding nvidia simply because of the fact that AMD actively supports the Open Source driver development. Sure, the Open Source driver does not have the performance issues fixed yet, but that is constantly changing and the driver is improving by leaps and bounds. If anything, if you have an ATI/AMD Graphics card and want to use wine, the best bet is to use the open source driver and keep a close eye on that driver.



Also, Look at Mesa with the Hyperz added for perf:
https://bugs.freedesktop.org/show_bug.cgi?id=36602

Well, I think and agree that using FOSS ATI drivers is the way to go on GNU/Linux, but I'd use FOSS drivers more often if:

- I had PROPER H.264 VA-API/VDPAU support. It's unacceptable that such a basic 2D feature is still unsupported by FOSS ATI drivers. (btw, VDPAU on MPEG2 videos already works well...)

The work on OpenCL that is going into the Mesa driver recently is actually key to this. If you have reasonable OpenCL and encoder/decoder that is implemented in OpenCL that will prefer the GPU over the CPU by default, you can expect the Video decoders to work properly.

- OpenGL games were not LIMITED BY THE CPU / MESA DRIVER, but rather BY THE GPU (on slow systems, you'll suffer performance penalties even if you use a High-End card).

I Can somewhat agree with this, but the MESA driver is improving. If anything I'd like reasonable OpenGL Performance on an athlon 64x2 (Socket 939 version, so ddr1 memory) with an radeon hd 4550 or radeon x1900gt.

- BETTER POWER MANAGEMENT (at least on laptop graphics cards). The only option we currently have if we buy a laptop with AMD/ATI graphics card is using it with Catalyst. Otherwise, you can expect to send you computer to the manufacturer before the guarantee expires (due to heating issues)...

Power management has not been an issue with the Radeon HD 6000 series ( A6 series cpu with a hd6650 mobile graphics card) because I am able to easily see about 4 hours on my laptop

With these news, I'm feeling a little more happier that I switched to Intel+nVidia (Bumblebee)... Maybe my next desktop system will also suffer a "change"...

Cheers

p.s.: I hope that Catalyst 12.5 (if it gets released) supports r600/r700-generation cards...

I Actually believe that the cards that where dropped with the 12.6 driver actually is a sign that the Open Source driver is actually getting enough steam. Sure the performance is not where it needs to be yet, and features are missing, but at least you are getting active support from AMD on this front ( initial support, and then of course the documentation for the various graphics registers for the card)... Actually, The only reason that the Open Source driver has poor performance and feature support is that there are not enough developers working on the driver actively. If anything, There is probably less than 10 developers actively working on the open source driver ( MESA and KMS ), and often times these developers are working on at least 5 different graphics cards at the same time.

[TobiSGD]

I would actually suggest avoiding nvidia simply because of the fact that AMD actively supports the Open Source driver development.

You are right. I can see the advances in open source drivers when they still not work correctly with hardware released 2008 (and still sold) when I can use Nvidias proprietary drivers with hardware released 2004, using the whole hardware and not only a part of it.

The work on OpenCL that is going into the Mesa driver recently is actually key to this. If you have reasonable OpenCL and encoder/decoder that is implemented in OpenCL that will prefer the GPU over the CPU by default, you can expect the Video decoders to work properly.

As seen posted in a different thread by bridgman, shader based video acceleration is NOT the solution. They stopped working on that and are working on UVD again, without knowing if they ever can release that code.

[QUOTEI Actually believe that the cards that where dropped with the 12.6 driver actually is a sign that the Open Source driver is actually getting enough steam. Sure the performance is not where it needs to be yet, and features are missing, but at least you are getting active support from AMD on this front ( initial support, and then of course the documentation for the various graphics registers for the card)...][/QUOTE]So dropping support for so called "legacy" cards, which surprisingly are the top of the line cards when it comes to integrated video solutions for chipsets for their top of the line CPUs and still are sold, and then putting that support into free drivers that only support half of the chip you just paid for and that with bad performance compared to the proprietary driver is a good sign? Then I would like to know what for you is a bad sign.

[evolution]

I would actually suggest avoiding nvidia simply because of the fact that AMD actively supports the Open Source driver development. Sure, the Open Source driver does not have the performance issues fixed yet, but that is constantly changing and the driver is improving by leaps and bounds. If anything, if you have an ATI/AMD Graphics card and want to use wine, the best bet is to use the open source driver and keep a close eye on that driver.

Well, I also don't like too much the fact that nVidia doesn't support Optimus/FOSS drivers on Linux, but the fact is that they seem to work at least at some extent (with Bumblebee). And a thing I've learned in engineering is that some things sometimes can be "beautiful" (such as the FOSS ATI driver project), but if they don't work as expected, they're (almost) useless.

Also, Look at Mesa with the Hyperz added for perf:
https://bugs.freedesktop.org/show_bug.cgi?id=36602

It seems the support is still buggy. Btw, my intel HD3000 IGPU already does that for quite a while... But maybe I'll give it a try on my r700 card when it gets more stable...

The work on OpenCL that is going into the Mesa driver recently is actually key to this. If you have reasonable OpenCL and encoder/decoder that is implemented in OpenCL that will prefer the GPU over the CPU by default, you can expect the Video decoders to work properly.

But it seems most of the work on OpenCL has been done on Evergreen-generation cards. Currently, I don't own any card from that generation (I've a r700-generation one). It seems OpenCL on HD4xxx series will be treated as a second-class citizen... OpenCL can be effective for encoding, but I'm afraid it might not work so well for decoding... I just don't know how, for instance, Intel can give a proper video-acceleration API with support for full-HD H.264 whereas AMD can't do the same with UVD...

Power management has not been an issue with the Radeon HD 6000 series ( A6 series cpu with a hd6650 mobile graphics card) because I am able to easily see about 4 hours on my laptop

For me, it might not be completly a "show-stopper", but it'd be nice to have an easier way of managing power management than having to write commands on the terminal each time we want to change radeon's power profile (I know how to do it with sysV systems, but currently, I'm already using systemd, so I can't use rc.local and I'm stuck with the maximum frequency from the "default" profile at boot). I also suspect the PM features are more optimized for all-AMD platforms...

For the "average-joe" Linux users, changing things on terminals is an easy way to make them return to WinBlows/Mac OS X (aka paid Linux)...

I Actually believe that the cards that where dropped with the 12.6 driver actually is a sign that the Open Source driver is actually getting enough steam. Sure the performance is not where it needs to be yet, and features are missing, but at least you are getting active support from AMD on this front ( initial support, and then of course the documentation for the various graphics registers for the card)... Actually, The only reason that the Open Source driver has poor performance and feature support is that there are not enough developers working on the driver actively. If anything, There is probably less than 10 developers actively working on the open source driver ( MESA and KMS ), and often times these developers are working on at least 5 different graphics cards at the same time.

We can consider it as "support", but at the moment it's not usable for the "average Joe" user... If AMD would really like to give a REAL effort on FOSS drivers, they should've focus more on FOSS AMD drivers and less on Catalyst, and give us some of the features I've mentioned in earlier posts... Why can't they provide a full-FOSS driver (that works as expected) like Intel does?

Cheers

[bwat47]

Power management has not been an issue with the Radeon HD 6000 series ( A6 series cpu with a hd6650 mobile graphics card) because I am able to easily see about 4 hours on my laptop

Power management with a dedicated mobile card is close to unusable in my experience though. My older laptop has a mobility hd2600 512mb. With the oss drivers, even with profile forced to low I get VERY high temps compared to catalyst, and very low battery life. The fan howls like a wounded bear if I so much as move a window. This is the main dealbreaker for me with the oss drivers.

[bridgman]

As seen posted in a different thread by bridgman, shader based video acceleration is NOT the solution. They stopped working on that and are working on UVD again, without knowing if they ever can release that code.

Actually my post got cut off somehow and I didn't have time to retype the whole thing. There are still some shader options which seem attractive but haven't been pursued yet (eg using compute shaders, which have lower overhead) but the timing seemed right to push ahead on UVD.

So dropping support for so called "legacy" cards, which surprisingly are the top of the line cards when it comes to integrated video solutions for chipsets for their top of the line CPUs and still are sold, and then putting that support into free drivers that only support half of the chip you just paid for and that with bad performance compared to the proprietary driver is a good sign? Then I would like to know what for you is a bad sign.

??? The idea of putting more work into the open drivers is to, in your words, "support the other half of the chip" (actually more like 10%) and improve performance. How can that be a Bad Thing ?