My horrible experience with ATI

[bridgman]

Bugs should be reported to ati.cchtml.com

We are gradually adding more "time-varying" information to the release notes. I agree that identifying kernels where problems are known at time of release would be a good next step.

Just to be clear though, our "support statement", at the front of the release notes, is that we support released versions of RHEL, SuSE and Ubuntu distros, subject to the other requirements below, ie if you are running a released version of RHEL but your kernel is older than 2.6 then the driver would not provide support.

[mtippett]

And where should bugs be reported so that they're noticed by the AMD/ATI-Developers? The xserver still can't be quit outside from X.

ati.cchtml.com is a suitable place for recording issues, capturing sufficient system information and reproducibility information provides useful information. We are aware of each bug that goes onto ati.cchtml.com, but obviously we don't generally respond or provide guidance on any timeline or priority for fixes.

Regards,

Matthew

[skywarp04]

What good does it do?

I have visited this site to report bugs and have watched many of them go unnoticed for a long time. I currently own a Radeon 3850 and it works great, in Windows. In Linux it may as well be an IGP. I use Linux everyday and have been for the past two years. I haven't had a stable release of fglrx on my system yet. Atleast not what I would ever deem stable. Something is ALWAYS broken. I'm not even asking for much. I just want simple 3D acceleration. I don't play games. I would just like to be able to have simple compositing desktop effects, smooth playing video, and no glitches that cause my system to lock up. And, yes I know it's fglrx locking my system up because if I don't use it everything works fine. I'm like everyone else, I'm tired of hearing excuses for stuff not working. I have a media center with an Nvidia card in it and it works almost flawlessly. No random screen glitches, no random lock ups or duplicating windows. I'm so tired of hearing the excuses that my next purchase probably will not be an AMD video card.

Report bugs you say? Where do you want me to start? I could go add five right now with the newly released 9.7 driver and I've only been using it for 2 hours.

[bridgman]

I have visited this site to report bugs and have watched many of them go unnoticed for a long time.

When you say "this site" do you mean Phoronix or the bug tracker ? Reporting a bug here won't get it in front of the developers.

Report bugs you say? Where do you want me to start? I could go add five right now with the newly released 9.7 driver and I've only been using it for 2 hours.

As long as you are not reporting bugs which are already in the system, please go ahead. If the bugs have already been entered and have not already been closed, please add your information to the existing tickets. If the bug description that matches your symptoms is generic ("black screen", "doesn't work", "sucks" etc..) then compare the steps to reproduce with yours and start a new ticket if it seems likely to be a completely different problem.

You may find that the open drivers plus a compositor like Metacity or KWin using XRender already meet your needs. Those compositors use the 3D engine but do not require hardware accelerated OpenGL.

[Markore]

With the specific exception of 3D apps requiring high levels of GL support, I believe all of the GPUs and usage scenarios were already supported as well as or better than fglrx before the parts were dropped from fglrx. For 3D-intensive users we recommended that they stay with their existing distro release until the open source driver framework catches up with their needs.

If you see any other cases where products or usage scenarios are not supported by the open drivers, please let us know. Analog tv outputs are still causing trouble, but AFAIK everything else is supported.

That is exactly the problem. You are stopping Linux development process by your action of not supporting newer Linux distributions until open source driver "catch up".
If I used to use minimal 3D abilities (enough for my usage) then if I install newer linux distribution, my card is unsupported there, because AMD says so.
Why don`t you make SHURE that OpenSource driver "catches up" right away, in all areas, including 3D? Maybe some features never worked actually?
when you drop support for an product, one and a half year after user was buying it?
Also what will we do with things that didn`t even work quite well
from the moment of start selling Integrated GPU in chipset
till the moment you "drop" users of AMD platform (Like HDMI and SVHS TV-Out on my integrated AMD motherboard)
You are basically saying, some product features of integrated
AMD graphics will NEVER be used by AMD drivers in its lifetime?

Wouldn`t there be nice to users of Chipset-integrated AMD graphics to STAY supported but with already available fglrx version (not newer than latest "supported" fglrx for that hardware) but continue to MAINTAIN IT
so user can happily continue to use older version where you freezed support for, in newer Linux distributions (Maybe even only Newer LongTermSupport distro versions).

And, to answer your conclusion from the start of the conversation:
I AM extrapolating level of suport for future AMD graphic products, by level of support that I would recieve in the lifetime of IGP AMD products we are using right now.
And we concluded that AMD will mostly support its integrated graphics for 1,5 to 2 years (from inception) and it will drop it.
Therefore, there are fat chances that I will push my customers to use AMD integrated graphics on large deployments (including Notebook AMD platforms, that is the same integrated problem).

Only way we as users could say we are granted for our choice of AMD products is that you continue to maintain _current latest version of driver_ for particular product in a manner that ensure that long-time users recieve proper support for Newer LTS linux distributions
OR make _Shure_ with your contribution that user`s experience will not go away with open source drivers, in the MOMENT you stop developing driver for a IGP/older card.
And for god`s sake, develop fully working drivers BEFORE card is realeased to the public in the first place.

[Markore]

Agree. IMO, this is a fraud, seriously.

Please clearly indicate the range of the support kernels.

Yes this in insane. It should be known what versions are supported And working on then flawlessly and what are not.

[mtippett]

Yes this in insane. It should be known what versions are supported And working on then flawlessly and what are not.

As has been communicated previously in this thread, we "support" distributions, not kernels. A kernel version is mostly meaningless since it represents the base code. It does not indicate which patches have been applied and neither does it represent which config options are present.

If we were explicit about which kernel versions were supported, we would then need to go the next level and determine which config options are supported and nto supported.

Your comment above above "flawlessly" is what is causing the delay. There have been some changes that have caused issues that have taken a lot longer to resolve - and some are still pending.

Historically we have caught up with kernels a release or two after the upstream has released. This maps to our standard development cycle - there is about a 6-8 weeks between when we close development and when the driver is released.

[bridgman]

That is exactly the problem. You are stopping Linux development process by your action of not supporting newer Linux distributions until open source driver "catch up". If I used to use minimal 3D abilities (enough for my usage) then if I install newer linux distribution, my card is unsupported there, because AMD says so.

The open source drivers already have more than "minimal 3D abilities" on all of the affected GPUs. What specific 3D functionality do you consider "minimal" but not available in the open drivers today ?

Why don`t you make SHURE that OpenSource driver "catches up" right away, in all areas, including 3D? Maybe some features never worked actually? when you drop support for an product, one and a half year after user was buying it? Also what will we do with things that didn`t even work quite well from the moment of start selling Integrated GPU in chipset till the moment you "drop" users of AMD platform (Like HDMI and SVHS TV-Out on my integrated AMD motherboard. You are basically saying, some product features of integrated AMD graphics will NEVER be used by AMD drivers in its lifetime?

If you mean "some product features on older hardware... will never be used by AMD-supplied drivers on Linux, only on Windows" that is certainly possible, although I expect we should be able to figure out the examples you listed.

Wouldn`t there be nice to users of Chipset-integrated AMD graphics to STAY supported but with already available fglrx version (not newer than latest "supported" fglrx for that hardware) but continue to MAINTAIN IT so user can happily continue to use older version where you freezed support for, in newer Linux distributions (Maybe even only Newer LongTermSupport distro versions).

If we weren't supporting open source graphics development as well it's possible we could do that. I don't think we can do both, and open source drivers will give you longer support and more features over time than legacy closed drivers.

And, to answer your conclusion from the start of the conversation: I AM extrapolating level of suport for future AMD graphic products, by level of support that I would recieve in the lifetime of IGP AMD products we are using right now. And we concluded that AMD will mostly support its integrated graphics for 1,5 to 2 years (from inception) and it will drop it. Therefore, there are fat chances that I will push my customers to use AMD integrated graphics on large deployments (including Notebook AMD platforms, that is the same integrated problem).

That's fair, but then please don't cherry-pick the data. Make sure you include the Xpress 200 (RS4xx) as well, which was launched in 2004 and was also supported right up until 2009. You need to understand that product can stay in the retail supply chain for many years, and there *are* going to be products which get EOL'ed while they are still on a store shelf. I can walk into a local computer store and buy cards based on GPUs we introduced in the late 90s, with no indication that they are anything but the current state of the art. One of the benefits of supporting open source development is that thos products can stay supported long after any vendor would have dropped proprietary driver support for them.

Only way we as users could say we are granted for our choice of AMD products is that you continue to maintain _current latest version of driver_ for particular product in a manner that ensure that long-time users recieve proper support for Newer LTS linux distributions OR make _Shure_ with your contribution that user`s experience will not go away with open source drivers, in the MOMENT you stop developing driver for a IGP/older card. And for god`s sake, develop fully working drivers BEFORE card is realeased to the public in the first place.

As far as I know the 9.3 drivers work fine on the newest LTS (8.04, right ?). For the last 18 months or so we have been offering Linux support for new GPUs at the same time as other OSes.