AMD Dropping R300-R500 Support In Catalyst Driver

[phoronix]

Phoronix: AMD Dropping R300-R500 Support In Catalyst Driver

Beginning next month with the Catalyst 9.4 release, support for the R300/400/500 generations of graphics processors will be dropped from AMD's mainline ATI driver. In a move they hope will allow them to focus their efforts on newer and upcoming graphics processors, the mainline Catalyst driver on both Linux and Windows will stop supporting cards older than the Radeon HD 2000 series. Linux customers affected will be encouraged to use their open-source driver stack (xf86-video-ati or xf86-video-radeonhd and Mesa) or stay with the Catalyst 9.3 driver.

http://www.phoronix.com/vr.php?view=13559

[TheK]

A bit early, or? Some weeks ago several shops still sold DX9 cards here and now no more fresh drivers? As long, as they at least bring updates for new API versions...

[hobbes]

I own a AGP X1600 pro!

I knew it this was coming! Second time for me...

This time at least I'll still have a better opensource driver to play with.

I've changed to the opensource driver almost a year ago, but its no way ready for many users.

I can't even playback a real 1280x720p x264 without a frame dropping here and there.

This is very bad, bad news.

It is becoming very clear (at least for me) that ATI opensource development is losing momentum.

Should I still believe it?

[aaaantoine]

Looks like my switching to xf86-video-ati after Arch Linux dropped support for Catalyst wasn't such a bad idea after all.

[pawstar]

This absolutely blows. My X1950XTX still does not work properly under Linux, I can't get a multimonitor setup going the way I need under Linux. I also have a X2400 series card for another monitor so what does that mean - now only one monitor will work with the official driver which is connected to the newer but weaker card while the rest are now dead !

On top of that my laptop also has a X1700 which is less than two years old. Are you saying it looses support too? Less than two years. I am furious. So what, when Winblow$ 7 comes along, it won't have drivers for it as well? I can't stand Vista but XP doesn't render text & GUI correctly on my internal screen (high dpi).

Just great, both Winblow$ and Linux won't have official support for my two main video cards.

I had enough of AMD/ATI dragging their feet. I was thrilled when AMD decided to open up the card specs and therefore continued to support them, but the mediocre driver that I had to put up with for the last 2 years and now this is the breaking point. If I could go back, I would buy NVidia all the way as I should have.

AMD/ATI I hope you get the message.

Signed,
Royally pissed off customer
:mad::mad::mad:

[nico342]

That so bad for "old" cards users. :( I own an 2400HD...pci.... I'm safe for now.... I hope that they will fix more bugs now .....

Annoying video tearing......

[yoshi314]

It is becoming very clear (at least for me) that ATI opensource development is losing momentum. well, it looks to me that fglrx development team is losing momentum - they want to focus on a smaller range of cards in order to keep up.

[hobbes]

well, it looks to me that fglrx development team is losing momentum - they want to focus on a smaller range of cards in order to keep up.

I was referring to the no extra-effort put into the opensource driver as said on the news.

Will this move generate greater benefits within the open-source ATI stack? It does not appear AMD will be ramping up on their open-source efforts. In fact, just this week the RadeonHD driver took a serious blow as one of the three Novell developers that were responsible for its development was laid off. The remaining developers are also facing shorter work hours.

if you put all news coming from AMD/ATI this just doesn't look good.

[grigi]

I'm more worried about power-management on my notebook :(
I have been quite happy with fglrx on my X1600 Mobility...

[Kjella]

if you put all news coming from AMD/ATI this just doesn't look good.

In all fairness to AMD, it was a Novell developer that was laid off. That's hardly something you can blame AMD for. But yes times are though all around so if the open source drivers are to survive it may need to become more community, less paid employees. I may be mistaken but I haven't got the impression there's hordes of willing driver developers that have gathered since AMD opened up the specs...

[karthikrg]

I think withdrawing support for R500 series cards is bad.. they are jus abt 2-3 years old now.. and i'm sure AMD's gr8 catalyst team is gonna continue 2 goof up wonderfully as they have been doing all along even after removing this support and "shrinking the codebase".. thank god i switched to a laptop with Nvidia 9600M graphics.. the prop. driver runs wonderfully without any hitches for me

[yoshi314]

In all fairness to AMD, it was a Novell developer that was laid off. That's hardly something you can blame AMD for.that makes you wonder whether fglrx team also got reduced, doesn't it?

[Jon4u!SvenJonsson]

As an owner of two r400 based cards (an x800gto agp and an x700 mobility in my laptop) i am not happy with amd's decision.

The current opensource driver (xf-86-video-ati and the mesa stack) is nowhere near fglrx in terms of stability and features (at least for me).
3D using the opensource drivers (eg for playing older games like doom3 or nwn) results in lockups or graphical corruption. Features like powerplay are not supported. Using the opensource-driver on my x700 results in white-screen on modeswitch / vt-switch. Using the opensource-driver on my x700 and my x800 with kde-4.2 results in lockups when desktop-effects are enabled.

I know that this is not amd's fault, and i also know that the opensource-developers work hard to improve the situation, but current situation is not good for owners of r400 based cards. There are too few opensource-developers working on the drivers, and the current focus seems to be more on bringing new features to the driver, then fixing bugs, resulting in an total loss of stability. The modeswitch for example worked great around xf-video-ati-6.8.0, but now with everyone switching to kms i frequently get a white-screen on my laptop. 3D worked somewhat (at least i could play nwn) until mesa-7.0, but now even starting glxgears results in an hardlock on my x800.

Bad times for ati-users i think.

A disappointed user.

cheers Jon

[karthikrg]

What is so stable about fglrx.. tearing videos with compiz&Xv + slow 2D (no render acceleration) + Heavy problems with wine.. are just some of the problems.. not to mention that for about 10 releases of fglrx from oct 2006 to aug 2007.. my x1800 never worked.. it just used to blackout when X used to start.. this is fglrx for u.. i have never had problems with xf86-video-radeon on KDE 4.2 personally.. infact it has been a smooth experience with compositing enabled also

[aradaj]

This news just sent shivers down my spine and I immediately remembered this little rant about Vista and it's adverse affects on linux in general.

http://www.cs.auckland.ac.nz/~pgut001/pubs/vista_cost.html

specifically the section about "Denial-of-Service via Driver/Device Revocation"

As I understand it, companies such as nvidia and ati are in effect forced to abandon UDA as decribed by the atricle linked above. And I believe that also has an adverse affect on linux because we are such a small market segment. I know that it also takes alot of time and resources to support legacy hardware, and with limited resources it doesn't make financial sense for ATI/AMD to keep supporting it.

I am a happy owner of a AIW-9800Pro, and recently an XFX HD4350 (to support open source development), and if ATI would have come out with the HD4850 6 Months earlier than when i bought my 8800gt, i would have been a happy owner of a hd4850 instead.

[Jon4u!SvenJonsson]

What is so stable about fglrx.. tearing videos with compiz&Xv + slow 2D (no render acceleration) + Heavy problems with wine.. are just some of the problems.. not to mention that for about 10 releases of fglrx from oct 2006 to aug 2007.. my x1800 never worked.. it just used to blackout when X used to start.. this is fglrx for u.. i have never had problems with xf86-video-radeon on KDE 4.2 personally.. infact it has been a smooth experience with compositing enabled also

Personal experience varies :).

For me fglrx worked well since around 2006 (and i used it since release 3.14, golden times back then :p ). I dont say that the opensource-drivers are bad per se, but for my situation they are unuseable. Currently the opensource-developers are focused on 1) supporting new cards and 2) supporting new features, both good things, but stability for older cards really sucks. Too few developers i think.

cheers Jon

[chrisr]

I bought my laptop in July 2007, and it was a "latest model" Lenovo T60p with a FireGL V5250. This chip is apparently akin to a R535, and it plays World of Warcraft acceptably under Fedora 10 at 1680x1050 resolution. At least, it does now. For many months after I first bought my laptop, the fglrx driver used to lock up when playing WoW. For months after that, it needed to pretend to have a X1700 chip instead. And now I learn that the Catalyst 9.3 release will be the last release for my laptop at all.

Gee, thanks.

Does Catalyst 9.2 support the Linux 2.6.28 kernel? Will Catalyst 9.3 at least support the 2.6.29 kernel? (I'm still using the 2.6.27.x series.) I bought this laptop specifically to be a portable "World of Warcraft machine" and the Mesa 3D support simply isn't ready to replace fglrx yet. (And that includes R300 chips too!)

[susikala]

Good move by AMD, kudos, there's little need for functionality duplication. Though I would completely stop developing Catalyst for Linux and concentrate the efforts purely on the FOSS drivers.

[[Knuckles]]

Well there goes support for both ATI cards I use... I think R500 is too soon, my local store still sells R500 cards!

I hope those drivers get pretty darn good.

[marcobrancalion]

AMD people want to force us to buy newer video cards, eh?.....
I got the message very clearly. A new nvidia card will be here soon ..... :D

[Nille]

This make me as ATI Costumer very angry. Drop all pre R600 is ok but why not maintain a legacy Driver with basic Kernel/X-server Support ?

I have an old R300 Card ( Radeon 9500 ) and this card dont work with the Open Source Driver ( the System freeze if i use XV or 3D )

AMD people want to force us to buy newer video cards, eh?.....
Nvidia too they relabeled the 8XXX to 9XXX.

I got the message very clearly. A new nvidia card will be here soon ..... :D

Nvidia is not Better and nobody know how long nvidia maintain the legacy driver pack.

[DragonsTear]

If you think about it it's a very clear and logic step.

Okay, 3D support ist still far away from being good - but in the moment, the development of the AMD FOSS drivers is more than just fast.

With 2.6.30, all AMD cards that exist now, will be supported...

I think it is a wise decision as AMD hasn't got any trump cards left - and the Catalyst driver is in a bad shape. Very very bad shape.

Switching to nvidia doesn't make sense to me - the nvidia drivers are in bad shape too - and there is no line on the horizon. Nouveau is far behind the ati development I think...

I hope of a much improved catalyst driver with version 9.4 ( otherwise I really start to hate the Catalyst Driver -.- ) ...

[shamooki]

Hey, what's that? I bought a T60 Thinkpad two years ago with X1300 mobility. But until last fall I was not able to use dynamic display switching in a comfortable manner! Only with the introduction of XRANDR this was useable.

And now they discontinue my chip!!!! And the OSS drivers don't support serious powermanagement; RadeonHD which I'm using at the moment doesn't even support automatic adjustment of display brightness when unplugging power. I can forget working on battery. Thanks a lot.

Oh man AMD/ATI, just when I was about to make friends with you again...

If I only listened to my gut feeling and got a machine with Intel graphics. You can bet that my next laptop will definitely come with Intel. Enough is enough.

[rvdboom]

Chill out, guys.
I can understand that some people who still have bugs in the fglrx driver and who cannot use the open-source one can be pissed off, but I guess most people using the fglrx driver now are with pre-r600 cards are satisfied with the 9.3 release, don't they? At least, that's what seems to come out from the above comments.
So just keep using the 9.3 release, what's the problem with that?
The whole X chain has been given a huge overhaul the last two years and it's not finished yet. It's no surprise people have difficulties to keep drivers going along while adapting to the new architectures. In a year or so, with Gallium3D, GEM and KMS setting down, I'm sure open-source drivers will become much more stable and less buggy as they will be able to rely on more solid ground.
I guess AMD is counting on that and thus is dropping support for older cards because they feel the opend-source driver should soon provide proper support. I may have done it reverse way, but, well...

[Ant P.]

Yep, this is all a huge knee-jerk reaction.

If AMD were retroactively removing R300 support from existing drivers, THEN you might have a case for all this complaining. 9.3 isn't suddenly going to stop existing.

[duby229]

This is great news. FGLRX should be dropped completely, but this is a good start. besides dropping fglrx completely, which likely wont happen, the only thing left is to consolidate the DDX drivers into a single project and to get galluim3d and gem up to par.

[yotambien]

I generally agree with the last comments. However, I own an R300 card, which is reasonably supported by the OSS driver and for which fglrx works well (at least the versions that work, heh). I understand that people with R500s are rather pissed off if their relatively new cards don't work very well with either driver.

The two main problems about this, as I see things, are:

-you will not be able to upgrade the X server and kernel if you decide to keep fglrx.
-laptop users have even less choice (if they're concerned about power management).

In the optimistic side, there seems to be a bright future for the open source drivers, which hopefully will satisfy most of us. It also remains to be seen whether dropping support for older hardware really makes things easier for AMD developers and fglrx quality for newer cards actually improves.

As a side thing, does anybody know why power management is rather poor in the open source drivers? Lack of man power or closed specs for those parts involved?

[Jon4u!SvenJonsson]

...

As a side thing, does anybody know why power management is rather poor in the open source drivers? Lack of man power or closed specs for those parts involved?

lack of manpower i think. there is a branch by agd5f which has basic powerplay support:
http://cgit.freedesktop.org/~agd5f/xf86-video-ati/log/?h=agd-powerplay

lack of manpower is also the reason why the opensource-drivers are far behind fglrx in terms of 3d-performance. and i doubt that this will change. i fear we will be waiting for gallium, memory-manager et al much longer then this year.

cheers Jon

[yogi_berra]

Yep, this is all a huge knee-jerk reaction.

If AMD were retroactively removing R300 support from existing drivers, THEN you might have a case for all this complaining. 9.3 isn't suddenly going to stop existing.

It is not as knee jerk as you'd like to think. As that video card will still be functional after xorg-xserver 1.7 is released but the 9.3 driver will not work with it and it is doubtful the OSS drivers will provide anywhere near the performance of the blob by that time.

Sure AMD is losing money but forcing everyone that uses their GPU's to upgrade is not the way to go about raising funds. These are the dumbass boardroom decisions that lose customers, not keep them.

[energyman]

This absolutely blows. My X1950XTX still does not work properly under Linux, I can't get a multimonitor setup going the way I need under Linux. I also have a X2400 series card for another monitor so what does that mean - now only one monitor will work with the official driver which is connected to the newer but weaker card while the rest are now dead !

On top of that my laptop also has a X1700 which is less than two years old. Are you saying it looses support too? Less than two years. I am furious. So what, when Winblow$ 7 comes along, it won't have drivers for it as well? I can't stand Vista but XP doesn't render text & GUI correctly on my internal screen (high dpi).

Just great, both Winblow$ and Linux won't have official support for my two main video cards.

I had enough of AMD/ATI dragging their feet. I was thrilled when AMD decided to open up the card specs and therefore continued to support them, but the mediocre driver that I had to put up with for the last 2 years and now this is the breaking point. If I could go back, I would buy NVidia all the way as I should have.

AMD/ATI I hope you get the message.

Signed,
Royally pissed off customer
:mad::mad::mad:

it doesn't 'loose support' The old drivers will continue to work. Or will the windows driver stiop working with the release of 9.5? No! Nobody forces you to update the drivers. And with win7 your hardware is unsupported anyway - so what are you complaining about?

And linux? Well - the opensource activities happened for a reason. You want better opensource drivers? Then sit down on your ass and start coding.

People like this make me sick.

[remm]

Ok, so:
- with all the architecture changes showing up, proprietary drivers have a tough time ahead, and unless AMD spent big time engineering on these old cards ...
- this will create more demand for the open source driver, which will recieve more testing and attention (Gallium has R300-R500 support under development right now)
- in concrete terms, you still get about 6 more months with the closed source driver, assuming you update your distro often
- radeonhd is a failed effort from day one (started behind closed doors by Novell; what a great plan ...), so would it be possible to stop mentioning it ?

But useless whining is good, this is the Internet, yah :)

[sundown]

It's funny. In a way they are stopping the support of something that wasn't quite supported in the first place. That stupid driver was always half-chewed, half something else and now they are throwing it up on us. That's right, that diriver was always a vomit.

[grigi]

It isn't made very clear.

Will the last drivers with R300-R500 support X.Org 1.6?

If no:
- That sux

If yes:
- That sux too, (although less)
- What if the last drivers has a problem which will essentially stop X.Org 1.6 from working? - No more support, means that we won't ever get it working.

So, I would like to find out the following: (Michael?)
Will the last drivers support X.Org 1.6?
If so, is there a chance of at least one more driver to fix any significant X.Org 1.6 compatibility bugs that may arise?

[Michael]

So, I would like to find out the following: (Michael?)
Will the last drivers support X.Org 1.6?
If so, is there a chance of at least one more driver to fix any significant X.Org 1.6 compatibility bugs that may arise?

AMD doesn't know the answer to it at this time. There's a chance the X Server 1.6 support may be back-ported, but I would not hold my breath. Don't count on any more legacy driver releases to "clean up" the support.

[grigi]

Thanks for the answer. It seems to call for a "wait and see" policy...
You will at least post it as as news item when it is confirmed either way?

[Michael]

Thanks for the answer. It seems to call for a "wait and see" policy...
You will at least post it as as news item when it is confirmed either way?

Sure thing.

[RealNC]

This has no real consequence on Windows since new Catalyst drivers didn't add anything new to X1000 cards and below. Quite the contrary; people were still running 7.x versions on those cards because those worked best (faster, more stable). I did too on my old X1950XT. Catalyst 7.11 was faster than anything newer and rock stable.

On Linux, however, this screws people. The Linux driver is bugged like hell and people were always waiting for new Catalyst releases to fix them, contrary to Windows where nothing really new was added and everything was working 100% in old drivers.

But, this has no consequence to AMD. Even if Linux people go NVidia now, AMD isn't affected much by it. They don't make money out of Linux people. Why should they consider us if there's no monetary gain for them in doing so? To sell 1 or 2 cards to Linux people while they sell 100 to Windows ones?

That's how it is. We don't have to like it, but that's the situation.

[NSLW]

In my point of view ATI can't manage to fix essential bugs with R300/R400/R500 so they drooped support for older Hardware.
Their developers are saying:
We aren't able to write good graphic driver and we should additional support something that we haven't ever properly supported? Let us make life easier and drop support for this Hardware, the buyers of R300/R400/R500 will understand it for sure.

[RealNC]

the buyers of R300/R400/R500 will understand it for sure.

Nope. It's more I like I wrote above: "who cares about the buyers of those since there's no income there."

[bridgman]

The current opensource driver (xf-86-video-ati and the mesa stack) is nowhere near fglrx in terms of stability and features (at least for me). 3D using the opensource drivers (eg for playing older games like doom3 or nwn) results in lockups or graphical corruption. Features like powerplay are not supported. Using the opensource-driver on my x700 results in white-screen on modeswitch / vt-switch. Using the opensource-driver on my x700 and my x800 with kde-4.2 results in lockups when desktop-effects are enabled.

I know that this is not amd's fault, and i also know that the opensource-developers work hard to improve the situation, but current situation is not good for owners of r400 based cards. There are too few opensource-developers working on the drivers, and the current focus seems to be more on bringing new features to the driver, then fixing bugs, resulting in an total loss of stability. The modeswitch for example worked great around xf-video-ati-6.8.0, but now with everyone switching to kms i frequently get a white-screen on my laptop. 3D worked somewhat (at least i could play nwn) until mesa-7.0, but now even starting glxgears results in an hardlock on my x800.

Jon, most of the current open source work is going into exactly the areas you mentioned, and the "new features" you're talking about are aimed primarily at those two areas -- kernel modesetting to provide a generic solution for system-specific VT-switch and suspend/resume issues, and rewriting the bottom end of the 3D stack to deal with ongoing performance and stability issues.

Gallium3D work for your GPUs has moved into Mesa master (rather than being a branch-off-a-branch, and I expect power management work will resume once some of the other projects rewriting the kernel driver settle down a bit. Bottom line, however, is that the level of activity on the open source drivers is higher than it has ever been before.

[RealNC]

It still seems progress is slow though. Do the devs working on then do so exclusively, or is it just one of the projects they're working on?

[Saist]

here's what I wrote on MepisLovers about this

March will be the last month that the R300 architecture, launched back in 2003 with the Radeon 9700, will be supported under AMD's Fglrx releases. As of the April Catalyst 9.4 set, all of AMD's OpenGL 2.0 cards will be removed from support, on both the Windows and Linux drivers. (so no, this isn't AMD picking on Linux, Windows is losing support too). These are all graphics cards from the Radeon 9500 - 9800 series, x300-x800 series, and x1x00 series.

Phoronix goes over how the news affects upcoming distributions : http://www.phoronix.com/vr.php?view=13559

The bad news is that AMD has no plans to update the final 9.3 driver to work against future X.org servers, a practice Nvidia carries out on the 71.86.xx, 96.43.xx, and 173.14.xx driver sets.

The good news is that improvements alread in development for the X.org Mesa drivers will probably bring the 3D support for R300-R500 GPU's in within 10% or so of the existing Fglrx driver, and 2D support is in many cases, faster on the existing Open-Sourced Drivers.

The other good news is that the decrease in product support will mean fewer regressions, and should simplify driver development for the RadeonHD series.


***

Okay... my take. I'm not really sure what to think. I'm a little uncomfortable... okay, a lot uncomfortable with the UAE taking an interest in AMD. If this change in support hadn't occured so shortly after the board of directors changed up, I'd be a little bit more willing to just pass it on as standard obselence.

Phoronix also makes mention that the hours had been cut on RadeonHD developers. However, what isn't mentioned is whether or not these shorter hours are by the request of AMD to Novell, or by Novell directly. What some people have also noted is that the X.org ATi driver has seen faster development than the RadeonHD driver, mostly due to Novell's approach being based on a complete reverse-engineering of the hardware and registers, rather than by using AMD tools such as AtomBIOS. Last year though, the RadeonHD developers pretty much gave in and started to use AtomBIOS : http://www.phoronix.com/scan.php?pag...atombios&num=1

Only time will tell if the new AMD after the split off of the Foundry, and the loss of Hector Ruiz to the new Foundry Company, will continue to support Open-Source... I'm just hoping that some of the moves over the past couple of weeks are just knee-jerk reactions by Novell and not indicative of a change in AMD...

otherwise, Intel's Larrabee.. and yes, this is ME saying this... Intel's Larrabee suddenly just a heck of a lot more attractive.

[sreyan]

@saist: RadeonHD also uses atom bios now for some cards

Honestly i assumed fglrx was not targeted at individual users but at corporations that need fast OpenGL support for workstation use.

Those corporations don't download the newest distro every six months. They use a stable old release. And when they purchase new hardware, they'd like that to get all of the attention. It's better ROI for AMD's driver budget in a lot of ways.

I've always been much happier with the quality of the open source drivers. Of course the performance isnt as high, but if that's the concern you can profile the problem and fix it yourself. It's not like you're being left in the cold like the NVIDIA users were when 96 stopped working with the latest xorg and 173 required SSE. At least the OSS ati drivers have more mature 3d.

[Saist]

@saist: RadeonHD also uses atom bios now for some cards

I know. I said that.

What some people have also noted is that the X.org ATi driver has seen faster development than the RadeonHD driver, mostly due to Novell's approach being based on a complete reverse-engineering of the hardware and registers, rather than by using AMD tools such as AtomBIOS. Last year though, the RadeonHD developers pretty much gave in and started to use AtomBIOS : http://www.phoronix.com/scan.php?pag...atombios&num=1

[sreyan]

I know. I said that.

yes you did :) apparently i can't read.

[yotambien]

I've always been much happier with the quality of the open source drivers. Of course the performance isnt as high, but if that's the concern you can profile the problem and fix it yourself. It's not like you're being left in the cold [...]

Really? What button do I press to fix them?

At least the OSS ati drivers have more mature 3d.

What's 'more mature'? And 'more mature' than what?

[bridgman]

It still seems progress is slow though. Do the devs working on then do so exclusively, or is it just one of the projects they're working on?

All three of our devs are full time on open source drivers, although Cooper's time is being split between new GPU support and ongoing support of Linux server OEMs with older GPUs (typically RV100-class parts).

The bigger issue is that most of the "thousand lines of code" work has been done and now we're spending more time in the "million lines of code" sections of the driver stack.

The other issue is that most of the community developer effort has shifted from fixing and tweaking the current drivers to implementing the next generation driver stack (KMS, GEM, DRI2, Gallium3D etc..). There's a ton of work being done but from an end-user perspective all you see is that the level of bug fixing seems to have dropped off.

The reality is that most of the remaining bugs aren't going to be fixable with tweaks and minor changes, and the dev focus is on the core issues that will provide more stable and performant drivers going forward.

[Saist]

All three of our devs are full time on open source drivers, although Cooper's time is being split between new GPU support and ongoing support of Linux server OEMs with older GPUs (typically RV100-class parts).

The bigger issue is that most of the "thousand lines of code" work has been done and now we're spending more time in the "million lines of code" sections of the driver stack.

The other issue is that most of the community developer effort has shifted from fixing and tweaking the current drivers to implementing the next generation driver stack (KMS, GEM, DRI2, Gallium3D etc..). There's a ton of work being done but from an end-user perspective all you see is that the level of bug fixing seems to have dropped off.

The reality is that most of the remaining bugs aren't going to be fixable with tweaks and minor changes, and the dev focus is on the core issues that will provide more stable and performant drivers going forward.

Okay. I think I'm taking this seriously out of context... but that really read something to the effect of: If Novell doesn't follow through on RadeonHD, our own guys are still writing the other driver.

I guess from my point of view, the loss of Novell as a backer really doesn't seem that important. As I understood the AMD strategy, the Novell developers just got an early lead on the same documentation everybody else would get at http://x.org/docs/AMD

[tball]

@Bridgman
Everytime I read something you post, I like what I hear. Really enjoy an Ati employed, posting in this forum.

You probably won't answer this, but I'll try anyway. Can you give a hint, how long the progress is with working 3d drivers for > R500?

[bridgman]

Okay. I think I'm taking this seriously out of context... but that really read something to the effect of: If Novell doesn't follow through on RadeonHD, our own guys are still writing the other driver.

Other driver... you mean -ati aka radeon ? We have been adding new GPU support to both drivers and I expect that will continue.

The bigger issue here, I think, is that most of the hard work on the DDX drivers (radeon and radeonhd) has been done other than bug fixing and ongoing addition of new GPU support. Most of the development efforts now are on the 3D side, in the drm and mesa drivers, and a lot of effort is going into moving functionality *out* of the DDX drivers.

There was an interesting milestone last night, when MostAwesomeDude got X running over the xorg state tracker, in turn running over Gallium3D (ie Gallium3D doing the basic 2D acceleration), running over KMS, GEM and DRI2 on an RV410. I don't get surprised often but that sure surprised me.

You probably won't answer this, but I'll try anyway. Can you give a hint, how long the progress is with working 3d drivers for > R500?

We're still hoping to have rough parity with 5xx by the end of April. We have glxgears running in house -- it still has some problems but I think we know the cause for each of the problems. Richard and Cooper are working on texture support now.

[oyvind]

Thank You for all the years of crap support for my mobile and unreplaceable graphics card in an otherwise very OK laptop. It is good to hear that it is finally being dropped, that will be better for all. Now I am suddenly out of options, and that's just fine and dandy.

fglrx is no more, since support will cease completely.
radeon has 2D corruption issues and 3D is shabby in comparison to fglrx
radeonhd has very slow 3D


Reason for bitterness: It has never ever really worked f***ing well ever since I bought this stupid bloody laptop. Always some issue (often of the show-stopping kind), and always some headache.

--
Sincerely,
ATI Mobility Radeon X1400 <-- What a joke.

[tball]

We're still hoping to have rough parity with 5xx by the end of April. We have glxgears running in house -- it still has some problems but I think we know the cause for each of the problems. Richard and Cooper are working on texture support now.


I like what I hear. Then we just need powermanagement, and opensource drivers will be running on my laptop :-)

[bridgman]

fglrx is no more, since support will cease completely.
radeon has 2D corruption issues and 3D is shabby in comparison to fglrx
radeonhd has very slow 3D


The radeon and radeonhd drivers use the same 3D stack, so 3D performance and features should be the same - if they're not please start a new thread and let's try to get that addressed.

There is a lot of work happening on 3xx-5xx open source 3D now -- MostAwesomeDude's Gallium3D work, airlied's radeon-rewrite effort to consolidate all the recent work (by airlied, glisse, nha and others) on memory management and command submission etc..

[gentoofu]

Yay, go MAD :).


I wonder if Arch Linux dropping support for Catalyst was a good idea. If releases will assumed to be faster by dropping legacy support, they should probably give it another chance maintaining it for cards >r500 (assuming that the drivers will get better).

But with this move, I'm glad xf86-video-ati will potentially have bigger user-base.

[GreatWalrus]

Great.... So is AMD/ATI going to do this again in the future? Are we going to be forced to upgrade our cards every 1.5-2 years in order to get the [current] benefits of the Catalyst driver? Really, the HD 2000's were released in 2007, that is not long ago at all. Not to mention, most who buy pre-built PC's or laptops usually get cards from many months or years before. For example, I bought my Dell laptop in May 2007, and they gave me a radeon xpress 1150 which was released in May 2006! On the other hand, I will be building my next computer and will then have more of a decision on how recent my card will be.

I don't currently use the ATI Catalyst driver and this card isn't so great enough that I would use it for gaming, so I will just stick to the xf86 ati driver. Of course, I don't know if I could even use Compiz again without fglrx. But as for the next card I will purchase, well, it depends on Nvidia's response to this news. Will they drop support for 'old' hardware? If not, I might consider changing my wish list.

[oyvind]

The radeon and radeonhd drivers use the same 3D stack, so 3D performance and features should be the same - if they're not please start a new thread and let's try to get that addressed.

There is a lot of work happening on 3xx-5xx open source 3D now -- MostAwesomeDude's Gallium3D work, airlied's radeon-rewrite effort to consolidate all the recent work (by airlied, glisse, nha and others) on memory management and command submission etc..

I gave the wrong impression about my opinions of the open source alternatives in my rant-post, I can see that. Let me just clarify, I really do appreciate the hard work that is being done there.

However, the new technologies are still some time away, and that leaves a gaping hole in the current situation, now that fglrx-support is gone in the near future. Add to that the fact that I've always had problems with this card, and now I'll just have to trade fglrx problems with radeon or radeonhd-problems instead. Considering that the latest Catalyst-releases have worked reasonably well, that means worse problems and not too great 3D performance (I consider corruption bugs truly showstopping). And by the time the issues are resolved, I have probably moved on hardware-wise. Just leaves some frustration. I once actually had a hope of an OK working Linux driver that performs well and doesn't have glaring bugs that always pops up to say hello, before the laptop becomes obsolete.

[bridgman]

Both ATI/AMD and NVidia drop support for old hardware -- it just happens to be our turn. There's probably someone at NVidia reading this and thinking "whew, now we can drop some old GPUs" :D

This case was a bit different from the norm because the 3xx-5xx architecture worked well for so many years, from 2002 through 2006, so in many cases the same driver code that supported 7 year old products also supported relatively recent products as well.

Compiz should run fine with the open source drivers on X1xxx and lower parts today, although there wasn't much success with the IGP (radeon express) parts until Alex was able to dig around in some of our internal documentation to figure out how some of the blocks should be programmed.

[Stormking]

Thank you ATI, for finally making clear that you are just a bunch of %$§%$§&! Thank you for failing to deliver a decent driver for five years, now. Thank you for showing what kind of support your customers can expect in the future.

Thank you for nothing but hours and hours of trying to get you piece of %$§/& working.

Don't worry, I will not bother you, again. Because I will never need a driver from you, again.

[lykowdk]

To the customer, this is a good move, permitting you are an owner of a Radeon HD 2000 graphics card or later (or an RS690/RS740 IGP, but the RS780 will remain supported).


Maybe it's because I went to public school.. What exactly is this line from the article saying?? Will the RS690 IGP still be supported or not? If not, how is it a good move for owners of a RS690 IGP?

Thanks

[Nicolas]

Thanks ATI.:mad:

Waiting since 2005 for a decent driver for my Mobility X700, and when you seem to be near to make basic acceleration to work without glitches, you drop support.

Your driver development strategy is really poor. I bet if I install my old Geforce4 Ti4400 and download the latest nVidia drivers it will work, and it's way older then my Mobility X700.

As a matter in fact, you say that nVidia also drops support on old hardware, but the difference is that even when using old drivers, they WORK. You can watch videos, use Compiz and do whatever you need to do.

You still have problems to play videos. Is not like I'm asking to be able to play FarCry2 using Wine while watching 1080p content. We want the basics to work.

Seriously... What's wrong with the decision makers on ATI?

[Saist]

Great.... So is AMD/ATI going to do this again in the future? Are we going to be forced to upgrade our cards every 1.5-2 years in order to get the [current] benefits of the Catalyst driver? Really, the HD 2000's were released in 2007, that is not long ago at all. Not to mention, most who buy pre-built PC's or laptops usually get cards from many months or years before. For example, I bought my Dell laptop in May 2007, and they gave me a radeon xpress 1150 which was released in May 2006! On the other hand, I will be building my next computer and will then have more of a decision on how recent my card will be.

I don't currently use the ATI Catalyst driver and this card isn't so great enough that I would use it for gaming, so I will just stick to the xf86 ati driver. Of course, I don't know if I could even use Compiz again without fglrx. But as for the next card I will purchase, well, it depends on Nvidia's response to this news. Will they drop support for 'old' hardware? If not, I might consider changing my wish list.

Thank you ATI, for finally making clear that you are just a bunch of %$§%$§&! Thank you for failing to deliver a decent driver for five years, now. Thank you for showing what kind of support your customers can expect in the future.

Thank you for nothing but hours and hours of trying to get you piece of %$§/& working.

Don't worry, I will not bother you, again. Because I will never need a driver from you, again.

Directed at both of you.

Nvidia has done this driver dropping thing for 3 releases now, .76, .96, and now 1.7x. They have ALREADY DONE THIS.

In addition, Nvidia doesn't even have an OSS driver for you to go to. So you have to either upgrade your Nvidia card to a newer version, or hope that Nvidia updates the legacy drivers, which could leave you months with no driver support.


By comparison, the last time ATi dropped a driver was back in 2006, and it was for an architecture from 2001.

So get over yourselves and quit trolling.

[kraftman]

I believe that also has an adverse affect on linux because we are such a small market segment. I know that it also takes alot of time and resources to support legacy hardware, and with limited resources it doesn't make financial sense for ATI/AMD to keep supporting it.

It looks windows is small market segment too:

...the mainline Catalyst driver on both Linux and Windows will stop supporting cards older than the Radeon HD 2000 series.

Thanks to Open Source drivers Linux is in far better position here :D I have no doubt, AMD is one of the most Linux friendly companies.

[Svartalf]

Both ATI/AMD and NVidia drop support for old hardware -- it just happens to be our turn. There's probably someone at NVidia reading this and thinking "whew, now we can drop some old GPUs" :D


Heh... I had the same line of thought there. :D


This case was a bit different from the norm because the 3xx-5xx architecture worked well for so many years, from 2002 through 2006, so in many cases the same driver code that supported 7 year old products also supported relatively recent products as well.


It's actually impressive that you had a lineup that could have been supported reasonably well for the span of 7 years. For those that're griping about this, it's been something that you HAD to know was coming. This is the reality of the proprietary driver scene- it's part of why I'm eager to see FOSS drivers picking up some speed on the AMD parts. AMD can't afford to keep supporting a chip design they no longer really sell to the public including OEMs.

It just means we all as a community need to pull the FOSS driver story more together on those chips a little quicker. That's all.

And for me in the medium term, it means my R420 and Rv515 boards will have to come out of the machines they're in if there's a single issue with the fglrx driver that has the support. My R600 board I bought recently has serious issues with user switching (The other Radeons don't...) using fglrx so if there's an issue they'll have to be stat-bagged and held in the hopes that the Gallium3D support gels quickly for them.

[GreatWalrus]

Directed at both of you.

Nvidia has done this driver dropping thing for 3 releases now, .76, .96, and now 1.7x. They have ALREADY DONE THIS.

In addition, Nvidia doesn't even have an OSS driver for you to go to. So you have to either upgrade your Nvidia card to a newer version, or hope that Nvidia updates the legacy drivers, which could leave you months with no driver support.


By comparison, the last time ATi dropped a driver was back in 2006, and it was for an architecture from 2001.

So get over yourselves and quit trolling.

I'm not trolling. I am contributing to the discussion - learn the difference.
And as noted by the last comment on page 6, Nvidia's "dropped drivers" continue to work and will most likely work for a while to come. On the other hand, ATI's dropped driver will stop working whenever there is a new kernel or xorg release.

[Stormking]

So get over yourselves and quit trolling.

For five years ATI promised a working driver (it never really did). Then they had their amazing new codebase that was supposed to do everything better (it never did) and now, after five years of dealing with their crappy &§%/, they just drop support for my hardware.

And getting upset over this confession of failure is trolling? Whats's next, are you coming over to actually spit in my face? Because that's what you're doing, verbally, right now.

[bridgman]

With respect, I don't think we ever promised anything except to keep improving Linux support, did we ?

The "amazing new codebase" was the 3D stack, not the rest of the driver and AFAIK that's the part everyone is upset about losing from fglrx, isn't it ?

We aren't dropping support for your hardware, as much as saying "from this point on we will be supporting your hardware via the open source driver, which we just spent a pile of time, money and effort on...".

[deanjo]

Directed at both of you.

Nvidia has done this driver dropping thing for 3 releases now, .76, .96, and now 1.7x. They have ALREADY DONE THIS.

In addition, Nvidia doesn't even have an OSS driver for you to go to. So you have to either upgrade your Nvidia card to a newer version, or hope that Nvidia updates the legacy drivers, which could leave you months with no driver support.


By comparison, the last time ATi dropped a driver was back in 2006, and it was for an architecture from 2001.

So get over yourselves and quit trolling.

Moving to legacy with nvidia does not mean drop of support. Nvidia still maintains drivers dating back to the TNT era (1998). Take a look at the driver release dates. This is something that ati has no intention of doing.

[remm]

Directed at both of you.

Nvidia has done this driver dropping thing for 3 releases now, .76, .96, and now 1.7x. They have ALREADY DONE THIS.

Obviously, these people don't know what they are talking about ...

More seriously, if you are interested in Linux and open source, then you should buy your hardware thinking about using it with open source drivers. Otherwise, there's really no point, you could use Windows with the same open source applications that you use on Linux. The only use of the blobs are for waiting for open drivers to get there when the hardware is new.

I expect Larabee will kill NV and AMD in the Linux market.

[Stormking]

With respect, I don't think we ever promised anything except to keep improving Linux support, did we ?
The problem is, it never did improve. *Something* was allways broken. Like the video overlay on tv-out, which did not work properly for 18 months. The inability to switch between text console and X. The system freezes when one simply wanted to log out. You just seem not to get how crappy fglrx is.

But those were the good times, right now it does not work for me at all. Again, you failed to even notice the problem for 3 months, than it was supposed to be fixed but wasn't and now you're officially dropping support.

[russofris]

Someone may wish to explain to AMD that x1000 is not a legacy architecture. Products are being released with this tech "today"

New HP product, just launched. By AMD's definition it's legacy before hitting the store shelves.

http://geizhals.at/at/a408280.html

Frank

[Stormking]

More seriously, if you are interested in Linux and open source, then you should buy your hardware thinking about using it with open source drivers. Otherwise, there's really no point, you could use Windows with the same open source applications that you use on Linux.
I've heart rumors about people who use Linux because they actually like it (or, less specific, Unix), not because of a particular application or because it's open source.

[Melcar]

Third party vendors often act independently to AMD official policy when it comes to GPUs and drivers.

[lykowdk]

Someone may wish to explain to AMD that x1000 is not a legacy architecture. Products are being released with this tech "today"

New HP product, just launched. By AMD's definition it's legacy before hitting the store shelves.

http://geizhals.at/at/a408280.html

Frank

The X1250 is an RS690 chipset.. The article mentions it, but it is ambiguous about its future. I am going to assume that the RS690 will still be supported since AMD has been pushing it as their embedded solution. ( I received a mailer about the Sempron 200/210u with a 690 yesterday. )

[drag]

And linux? Well - the opensource activities happened for a reason. You want better opensource drivers? Then sit down on your ass and start coding.


Actually I would not think it's coding what is needed (well that is needed, but end users can't typically code). Anyways most people do not have the ability to do that.

What is needed from most people is going to be feedback.

For example:

Say you have a video card that freezes when you try to do 3D. Who is going to help you fix it if you sit on your hands and don't file reports or do anything about it?

NOBODY.

NOBODY IS GOING TO FIX ANYTHING IF THEY ARE UNAWARE OF IT.


---------------------------------


What is needed is the ability for people to run lots of tests on different hardware with the same exact software setup. So that developers can know easier which hardware is the problem and stuff like that.

They don't even need to be particularly good bug reports. Numbers make up for it.

So say you get 300 bug reports from different people. They are shitty bug reports, but you know the user and the hardware. Well if 200 of those people are having problems with the same chipset on their cards... then that helps to narrow down the problem significantly, right?


SOOOOO.....

What I am getting at is this:

Is there any Live Linux distribution out there right now for testing video hardware?

I want a Live Linux cdrom that is built with the latest and greatest open source drivers and comes with a bunch of benchmarking and hardware testing tools installed. So that a individual can download the image, drop it on a USB key or on a CDROM and then just engage in a bunch of testing.

Something quick and easy that does not require a bunch of Linux skills or require installing software over a existing system. Something even a typical Windows user can handle.


If that does not exist then I want to make one. I want to take Phoronix's benchmark tools and make a live CD/USB Key with the latest and greatest X Windows and OpenGL drivers.

[energyman]

Stormking - you want drivers? Great - sit down and start coding! The documentation is there. AMD/ATI put everything on the table you need to improve the FOSS drivers. I am expecting you sending patches upstream soon.

@the rest of the whiners. You act like AMD is activating a secret switch that will turn off your cards at the first of may. Nobody forces you to upgrade to a new X. And new kernel support can be patched it. Oh - and 2.6.27? It is planned to make it a long time supported kernel (maybe 2.6.28 too because of ext4) so you will have bug fixes for a loooong time. Just look at 2.6.16 how long that beast stayed around. All you loose is the ability to use the latest X with fglrx. So what? Instead of whining like little girls you should man up. For years people rambled that amd should just release docs, the community would do the rest, fglrx not needed after that. So you have the docs. Now stand by your words.

[bulletxt]

Ok, now I can finally say goodbye to AMD after this move.

Not only NVIDIA beats you 100 times, they even keep a legacy driver updated. This is enough. I'm sick of AMD and their shit.

If you use Linux, use NVIDIA cards.
I'm already putting my ATI 2600XT card on ebay even though AMD Catalyst shit driver supports it. I have enough of this cause I even own a 9600XT card. You wanna fool me? Fine, I won't give you my money anymore.

[Melcar]

Actually I would not think it's coding what is needed (well that is needed, but end users can't typically code). Anyways most people do not have the ability to do that.

What is needed from most people is going to be feedback.

For example:

Say you have a video card that freezes when you try to do 3D. Who is going to help you fix it if you sit on your hands and don't file reports or do anything about it?

NOBODY.

NOBODY IS GOING TO FIX ANYTHING IF THEY ARE UNAWARE OF IT.


---------------------------------


What is needed is the ability for people to run lots of tests on different hardware with the same exact software setup. So that developers can know easier which hardware is the problem and stuff like that.

They don't even need to be particularly good bug reports. Numbers make up for it.

So say you get 300 bug reports from different people. They are shitty bug reports, but you know the user and the hardware. Well if 200 of those people are having problems with the same chipset on their cards... then that helps to narrow down the problem significantly, right?


SOOOOO.....

What I am getting at is this:

Is there any Live Linux distribution out there right now for testing video hardware?

I want a Live Linux cdrom that is built with the latest and greatest open source drivers and comes with a bunch of benchmarking and hardware testing tools installed. So that a individual can download the image, drop it on a USB key or on a CDROM and then just engage in a bunch of testing.

Something quick and easy that does not require a bunch of Linux skills or require installing software over a existing system. Something even a typical Windows user can handle.


If that does not exist then I want to make one. I want to take Phoronix's benchmark tools and make a live CD/USB Key with the latest and greatest X Windows and OpenGL drivers.

That wouldn't be a bad idea. I'm all for testing drivers (which I already do), but having to build them in order to run the latest code can be a pain. If users can get a hold of a small image that is constantly updated with the major changes in open source drivers, I think it would provide greater incentive for people to become sort of beta testers for the drivers. After all, people want functional drivers (I would hope).

[energyman]

Ok, now I can finally say goodbye to AMD after this move.

Not only NVIDIA beats you 100 times, they even keep a legacy driver updated. This is enough. I'm sick of AMD and their shit.

If you use Linux, use NVIDIA cards.
I'm already putting my ATI 2600XT card on ebay even though AMD Catalyst shit driver supports it. I have enough of this cause I even own a 9600XT card. You wanna fool me? Fine, I won't give you my money anymore.

good bye and good luck. You will be angry again when others enjoy kms and you are stuck with nvidia.

[Saist]

good bye and good luck. You will be angry again when others enjoy kms and you are stuck with nvidia.

nah. He'll be angry when he finds out that Nvidia's drivers have more bugs than a Louisiana swamp.

[SyXbiT]

the decision is obvious
if you want to HOPE for a future of better drivers, improved FOSS, then buy ATI

if you just want stuff to work NOW, buy nvidia

I know what I want, and it's in the present.

seriously, who cares about KMS, it happens on bootup, that's it

[bulletxt]

good bye and good luck. You will be angry again when others enjoy kms and you are stuck with nvidia.

Lol if you really believe AMD will become the leader in GPU Linux market then fine, you're free to believe it.

If you look at history (including today), there is not even 1 reason to believe it considering NVIDIA continuous work. There is a reason if NVIDIA is the leader in the Linux market since year 2001. But no, you guys are too blind to listening at AMD fairy tales.

[pawstar]

it doesn't 'loose support' The old drivers will continue to work. Or will the windows driver stop working with the release of 9.5? No! Nobody forces you to update the drivers. And with win7 your hardware is unsupported anyway - so what are you complaining about?

What makes you say my hardware is unsupported? I've got equipment which can run vista which is far more resource intensive than Win7. The only problem will be the video card. You just can't expect one to go and buy a new one after 2 years of just spending more than $500 on one can you?

The current drives do not work - they never had under linux, so they will not continue to work - where is my working video overlay, where is working rotation, where is working multi-monitor and above all where is the stability. So many bugs with the later r500 models (i.e. what I have) that its not even funny. Coincidentally, 9.2 borked and I had no video output on my system. I had to go back to 9.1.

I cannot and will not remain with an old os - and why should I. It DOES LOOSE SUPPORT because newer Xorg and Kernel upgrades will bork the driver. Maybe you can't seem to understand that some of us need the new features that come out - even be it just stabilized support for other hardware in later releases of the kernel.

And linux? Well - the opensource activities happened for a reason. You want better opensource drivers? Then sit down on your ass and start coding.
Why should I? I paid good money for a card and I expect the company who got my $$$ to support the product they sold me. I've got far more important things to do than to screw around with writing my own drivers.

I've been hearing about the promise of opensource drivers for too long. That carrot-on-a-stick routine is way too tiring now. Already waiting 2 years and now told to wait even more for a working driver is not acceptable business practice IMO.

[pawstar]

Someone may wish to explain to AMD that x1000 is not a legacy architecture. Products are being released with this tech "today"


Dead on! I know the laptop that I have with the X1700 video card continued to be on sale in the mainstream until less than a year ago. And the video card is not upgradeable. After spending that much money on it, you can't just throw it out because of some stupid driver issue with a video card and spend a few g's again. That would be insane especially in this crappy economic climate..

[Qaridarium]

on my system the Beta of the Catalyst 9-3 ( 8.591) goes Rockstable!

:~$ infobash -v3
Host/Kernel/OS "ass" running Linux 2.6.28-1-686-bigmem i686 [ sidux 2008-04 Πόντος - kde-lite - (200812222321) ]
CPU Info (1) AMD Engineering Sample 1024 KB cache flags( sse3 nx lm svm
Videocard ATI RV710 [Radeon HD 4350] X.Org 1.4.2 [ 1920x1200@60.0hz ]
Network cards 2x Broadcom NetXtreme BCM5754 Gigabit Ethernet PCI Express
Processes 143 | Uptime 6:24 | Memory 423.6/8116.5MB | HDD SAMSUNG SP2014N,ATA SAMSUNG HD103UJ Size 1200GB (93%used) | GLX Renderer ATI Radeon HD 4350 | GLX Version 2.1.8543 Release | Client Shell | Infobash v3.05

my othe PC with an R300 aka X800 Runs well to! for now and in the future Opensource will rock da house :)

[bridgman]

What makes you say my hardware is unsupported? I've got equipment which can run vista which is far more resource intensive than Win7. The only problem will be the video card. You just can't expect one to go and buy a new one after 2 years of just spending more than $500 on one can you?

Microsoft usually maintains ABI compatibility across at least a couple of generations of OS (ie several years). Vista can normally work with XP video drivers, for example. Linux changes the API and ABI exposed to binary drivers with almost every new kernel release.

The current drives do not work - they never had under linux, so they will not continue to work - where is my working video overlay, where is working rotation, where is working multi-monitor and above all where is the stability. So many bugs with the later r500 models (i.e. what I have) that its not even funny. Coincidentally, 9.2 borked and I had no video output on my system. I had to go back to 9.1.

The 5xx generation of GPUs don't have the kind of video overlay you remember from earlier hardware. It's not a question of driver support.

Rotation and multi-monitor should both be supported in the open drivers today.

My impression was that stability with the open drivers was very good today; have you tried them recently ?

I cannot and will not remain with an old os - and why should I. It DOES LOOSE SUPPORT because newer Xorg and Kernel upgrades will bork the driver. Maybe you can't seem to understand that some of us need the new features that come out - even be it just stabilized support for other hardware in later releases of the kernel.

Again, the open drivers not only evolve with the new distros but they are often used to create the new features you are looking for.

I've been hearing about the promise of opensource drivers for too long. That carrot-on-a-stick routine is way too tiring now. Already waiting 2 years and now told to wait even more for a working driver is not acceptable business practice IMO.

Just curious, what are you waiting for from the open drivers ? AFAIK for your card (I'm guessing high end 5xx ?) they have caught up with fglrx in every area except gaming performance already.

Dead on! I know the laptop that I have with the X1700 video card continued to be on sale in the mainstream until less than a year ago. And the video card is not upgradeable. After spending that much money on it, you can't just throw it out because of some stupid driver issue with a video card and spend a few g's again. That would be insane especially in this crappy economic climate..

I don't think anyone is saying "throw the card out", are we ? We're just saying "from this point on we're going to be delivering support via the open source drivers".

[Ant P.]

There is a reason if NVIDIA is the leader in the Linux market since year 2001.

The same reason Windows XP is the leader in the OS market since 2001 - inertia. nVidia put out a working driver at a time when nothing else worked.

Also I should point out here that I'm using an ATi card from that era with more effects than Vista can crap out, whereas I've got an almost equal spec nVidia card gathering dust in a drawer, because they added that infamous black window bug then dropped support days later. Funny how compiz ran fine on it at one point, but that's forced obsolescence for you.

[Melf]

Well, this is an interesting coincidence :)
I have just switched to the open source radeon driver on my laptop(ubuntu 8.10, asus f3jr, mobility radeon x2300 = RV535(?)).
I switched because of stability problems with the catalyst driver(fullscreen video in through xv in compiz would lead to hard freeze).
This are my impressions so far:
+its very stable, suspend works, hibernation thorough s2disk too, everything while running compiz
+video is without ANY flickering in compiz, xv has good antitearing support and 720p video is watchable
-performance as measured by glxgears is roughly cut in half in comparison to the catylst driver(1100 vs 2000+) but the number is the same with and without compiz(on the catalyst, it was always lower with compiz)
-some 3D aps dont work :)
-problems with some aplications that run 3D in fullscreen in compiz,
looks like if scaling is not implemented (ap. wants to run 800*600 in fullscreen but it appears on a black screen in a box) and sometimes things *behind* a fullscreen aplication (gnome resource monitor etc.) flicker through,it can be solved by running the aplication in a window or without compiz (looks like without compiz the scaling works ok)
-the power management is probably not working, but I have not noticed any bigger difference, maybe it wast working before at all :D
-there is no setup utility, you can get info from xorg.log and settings go as usual to xorg.conf when neaded
-some minor corruption from timer to time but nothing showstopping

I have tried the radeonhd driver too but I got just 300fps in glxgears, maybe I forgot to add some options to xorg.conf ?

All in all I am quite pleased. I finally managed to achieve my main obejective, to have:
*stability
*suspend,hibernation
*good video playback
*compiz
all at the same time.
I would say that the open source radeon driver is not bad at all!
There are Ubuntu packages with radeon 1.6.11 and radeonhd 1.24 available so you can try the newest oss drivers yourselve.
https://launchpad.net/~tormodvolden/+archive/ppa
These packages are unofficial and HIGHLY experimental, so be carefull :)
(Text console boot options and ssh access may come in handy :)

[kraftman]

Obviously, these people don't know what they are talking about ...

More seriously, if you are interested in Linux and open source, then you should buy your hardware thinking about using it with open source drivers. Otherwise, there's really no point, you could use Windows with the same open source applications that you use on Linux.

There are hundreds more reasons why I use Linux instead windowses, or macoses.

The only use of the blobs are for waiting for open drivers to get there when the hardware is new.

Not exactly, but I usually choose Open Drivers instead of closed blobs.

I expect Larabee will kill NV and AMD in the Linux market

What's Larabee?

@Ant P

The same reason Windows XP is the leader in the OS market since 2001

What reason? Linux worked better for me those times and it works far better for me right now. The reason is ms monopoly.

@SyXbiT

seriously, who cares about KMS, it happens on bootup, that's it

It's much more.

[Stormking]

Stormking - you want drivers? Great - sit down and start coding! The documentation is there. AMD/ATI put everything on the table you need to improve the FOSS drivers. I am expecting you sending patches upstream soon.
Why would I? I already paid ATI to do that for me.

PS: Calling your customers - who just want what they paid for - "whiners", clearly is a statement of its own.

[bridgman]

mobility radeon x2300 = RV535(?)).

I have tried the radeonhd driver too but I got just 300fps in glxgears, maybe I forgot to add some options to xorg.conf ?

AFAIK the X2300 is RV550, more like an RV515 with an early version of UVD.

The radeon driver turns DRI on by default but radeonhd defaults to turning it off.

If you add :

Option "DRI"

to the Device section of xorg.conf that should give you the same performance in radeonhd.

EDIT - agd5f just mentioned in IRC that while radeonhd and radeon both have tear-free Xv code for 6xx/7xx, radeonhd probably doesn't have tear-free for 5xx yet.

[duby229]

With respect, I don't think we ever promised anything except to keep improving Linux support, did we ?

The "amazing new codebase" was the 3D stack, not the rest of the driver and AFAIK that's the part everyone is upset about losing from fglrx, isn't it ?

We aren't dropping support for your hardware, as much as saying "from this point on we will be supporting your hardware via the open source driver, which we just spent a pile of time, money and effort on...".

Which is exactly what you should be doing so thank you very, very much.

For those of you that are actually complaining, here are the facts....
1: The user base for the open drivers is about to grow substantially.
2: Many of those users are bound to be programmers, primarily due to the fact that Linux tends to attract that sort of crowd.
3: Many of those programmers are going to find problems with the drivers as they are today, because well, they are incomplete.
4: Many of those unsatisfied programmers are going to do what they can to get the driver working best for them, and that helps you.

Open source development is primarily a process of elimination. This is its strength and its weakness. Strength because it facilitates a --MUCH-- larger developer base then any proprietary effort could. These developers that will be fixing there own problems couldnt do the same using a closed driver. And while they're doing it, many other people will benefit.

The goal behind open development is to organize (Organize being the key word here) as many developers working together as possible. The only effective way of increasing the base of developers is to increase the user base substantially. This effort is going to do exactly that.

[elsie]

All in all I am quite pleased. I finally managed to achieve my main obejective, to have:
*stability
*suspend,hibernation
*good video playback
*compiz
all at the same time.
I would say that the open source radeon driver is not bad at all!

I just went to look at the status of the radeon driver for r500 series & you're right, it looks really good! I'd switch drivers right now if powerplay was working.

That puts my mind at ease.

[Michael]

Actually I would not think it's coding what is needed (well that is needed, but end users can't typically code). Anyways most people do not have the ability to do that.

What is needed from most people is going to be feedback.

For example:

Say you have a video card that freezes when you try to do 3D. Who is going to help you fix it if you sit on your hands and don't file reports or do anything about it?

NOBODY.

NOBODY IS GOING TO FIX ANYTHING IF THEY ARE UNAWARE OF IT.


---------------------------------


What is needed is the ability for people to run lots of tests on different hardware with the same exact software setup. So that developers can know easier which hardware is the problem and stuff like that.

They don't even need to be particularly good bug reports. Numbers make up for it.

So say you get 300 bug reports from different people. They are shitty bug reports, but you know the user and the hardware. Well if 200 of those people are having problems with the same chipset on their cards... then that helps to narrow down the problem significantly, right?


SOOOOO.....

What I am getting at is this:

Is there any Live Linux distribution out there right now for testing video hardware?

I want a Live Linux cdrom that is built with the latest and greatest open source drivers and comes with a bunch of benchmarking and hardware testing tools installed. So that a individual can download the image, drop it on a USB key or on a CDROM and then just engage in a bunch of testing.

Something quick and easy that does not require a bunch of Linux skills or require installing software over a existing system. Something even a typical Windows user can handle.


If that does not exist then I want to make one. I want to take Phoronix's benchmark tools and make a live CD/USB Key with the latest and greatest X Windows and OpenGL drivers.

We will have an answer for you needs soon :D

[Craig73]

Wow... so much negativity here. While I understand that people are frustrated at AMD for the problems each person has had with their drivers... there seems to be no recognition from what I understand are real challenges supporting the quirks of various distributions, kernels, and a changing graphics stack. Really for all the architectural issues people are trying to address in the rest of the stack, it's impressive what they do put out. (Windows, with it's slow rate of change and single distribution (sort of) seems like a dream).

And what is AMD's reaction to this landscape - they open their documentation and produce an open source driver. OK - it's not perfect yet, but they have grabbed onto philosophy that many feel is the better way to go, and are making that the key driver. And as the Linux landscape evolves... the open source driver should evolve right along with it.

Does it make more sense to fix the old driver? or work on the open source one and get it up to snuff sooner (which, it sounds like it is doing pretty good)

[Kano]

As they do not provide updates for legacy drivers this will lead to endless user supported kernel patches for 9-3 driver. You can guess that i really hate that. If you do a Debian Lenny based distribution running latest kernel and fully without ATIs support that a dramtic issue what they are doing. Except Xserver 1.6+ support for 71.xx driver series Nv supports all GPU - maybe not with added extra drivers. But those are old DX7 cards, R500 is a DX10 card and they drop support for everything below a DX10.1. Somebody should _sue_ ATI, really thats unfair if you can not use a newer Xserver, DRM, Mesa on a specific distro. The driver is not even in a fully working state and then drop support? Thats absolutely ridiculous!

[Saist]

Wow... so much negativity here. While I understand that people are frustrated at AMD for the problems each person has had with their drivers... there seems to be no recognition from what I understand are real challenges supporting the quirks of various distributions, kernels, and a changing graphics stack. Really for all the architectural issues people are trying to address in the rest of the stack, it's impressive what they do put out. (Windows, with it's slow rate of change and single distribution (sort of) seems like a dream).

And what is AMD's reaction to this landscape - they open their documentation and produce an open source driver. OK - it's not perfect yet, but they have grabbed onto philosophy that many feel is the better way to go, and are making that the key driver. And as the Linux landscape evolves... the open source driver should evolve right along with it.

Does it make more sense to fix the old driver? or work on the open source one and get it up to snuff sooner (which, it sounds like it is doing pretty good)


I know how you feel. What follows is my flame-off on the author of the SMXI / SGFXI script after listening a whine fest of his over on Mepislover : http://mepislovers.org/forums/showthread.php?p=161994#post161994

so in other words : whine whine, fuss fuss, it's always somebody else is at fault.

If you expected the worst, and did nothing to prevent that worst from occuring, either by getting involved or contacting AMD directly, what you say comes off as pathetic excuses, and nothing more.

***

okay. edit. After walking away from the keyboard for a while, this just came off as incomplete. I'm going to try to articulate what exactly it is about h2's stance that just completely ticks me off.

For starters, it's the baldface stance on the status of the drivers. Okay, at one point the OpenGL drivers were horrible. AMD/ATi fixed the acceleration problems. The list of bugfixes with each driver went from one or two bugs, to five or six, to seven or eight, to ten or more bugs squashed on average.

However, that wasn't good enough for people. Open source users cried for an open source strategy. So ATI, then AMD came up with one, releasing internal code documentation and paying for the development of an Open Licensed driver, RadeonHD.

Again, people complained because RadeonHD was competing with X.org ATi, and X.org ATi was seeing faster development due to their adoption of AMD tools, while RadeonHD went after the less AMD dependent reverse-engineering method. So the RadeonHD devs finally started using AtomBIOS.

That still wasn't good enough. People wanted parts of Fglrx opened up, so AMD said yes. Then people wanted the installer scripts opened up, so AMD said yes.

And now that AMD has pretty much answered every single question, rewritten the drivers from scratch, and damn near hand catered to every single demand brought forward from the Open-Source communities, it's still not enough. After distribution maintainers railed against AMD/ATi for lack of control over the scripts, and the inability to add more scripts for each distribution, now it's suddenly the opposite. AMD isn't doing enough internal testing. It's no longer the distribution maintainers problem. It's AMD's problem.

That's bull****, and it ends here, and it ends now.

[susikala]

Let me add something here. About 70% of this thread is people whining about this step. It gets kind of tiring after a while to read all the negativity.

Basically, what bridgman keeps saying here holds true. The FOSS drivers will rather soon (once the new graphics stack stabilises) reach about 60-70% performance of fglrx, which is _more than reasonable_ for something that is maintainable for a very long time. For the remaining 30-40% for your gpu you'd have to roll up your sleeves and get to work -- which is again, more than fair.

I mean, this is becoming ridiculous. The specs are there for pretty much everything and for everyone to see. When they weren't, people whined about AMD. Now people (even if it's technically a different group of people) still whine about AMD doing the _right_ step to be able to deliver overall better performance with fglrx (even though whether it's needed anymore is questionable). All in all, can't you appreciate AMD's efforts?

I mean, saying "my next card is nvidia" just doesn't cut it, if you work according to that logic, you might as well be using Windows.

[deanjo]

I mean, saying "my next card is nvidia" just doesn't cut it, if you work according to that logic, you might as well be using Windows.

Expecting on par performance levels of the card when in another OS in not unreasonable at all. People don't purchase video cards for 70% of it's potential. People will buy and upgrade cards when there is a 20% increase in performance though.

[bridgman]

R500 is a DX10 card and they drop support for everything below a DX10.1.

R500 is DX9 -- DX10 started with R600.

AFAIK the X1xxx parts are all DX9, HD2xxx are all DX10, HD3xxx and above are all DX10.1.

Somebody should _sue_ ATI, really thats unfair if you can not use a newer Xserver, DRM, Mesa on a specific distro. The driver is not even in a fully working state and then drop support? Thats absolutely ridiculous!

Again, we aren't dropping support we are shifting our support to the open source driver.

It would all seem more clear if we had hired open source developers today rather than starting a year ago, but we didn't want to shift support to open source until the drivers were in reasonably decent shape (which, in fairness, I think they are for 3xx-5xx GPUs).

[Qaridarium]

But those are old DX7 cards, R500 is a DX10 card and they drop support for everything below a DX10.1. Somebody should _sue_ ATI,

you are wrong! R500 was an DX9 cart!

First DX10 cart was the R600 hd2900!!! and thats DX10 not 10.1!

First DX10.1 cart was the 3XXX carts!

a R500 like the X1950 only has DX9!



somebody should sue you becourse of your lies!