Slashdot Mirror
Problems With OEM ATI Cards And ATI's Linux Driver
Doug Bostrom writes "Over at FlightGear.org, Andy Ross describes how ATI's new Linux drivers only seem to work with "official" ATI cards (made by ATI), why that does not make sense, and a possible fix that unfortunately would mean booting Windows, if only for a few minutes."
There are a few possible explanations for this rather odd driver situation...
I know this is rumor-mongering, but I can't help but notice that the *Windows* drivers dont' perform such a check, and neither do the Linux Retail drivers...
Consider this: Microsoft or some other party requests unofficially that ATi *not* support Linux in its OEM hardware, just for the sake of not having OEM desktop vidcard support for Linux...this could explain things like the OEM/Retail check that occurs in Linux, but not Windows. Interesting stuff..I want to see what ATI's reaction on this is.
------- "From bored to fanboy in 3.8 asian girls" ----------
This is probably a very good argument for non-binary, truly open-source drivers...
/* affect != effect */ void affect(int *thing,int effect) { *thing += effect; }
I think it was just a programer doing a standard check (helo vidocard. who are you?) and a QA department with every and only ATI cards.
Not so, not so, not so. ATI has a reason for ensuring that their drivers function properly only with authorized hardware. ATI's marketing strategy centers around the company being recognized for making the top-quality graphics cards on the market. This definition includes all components from circuit boards to microchips. ATI's primary market is those consumers who need or want top-of-the-line video cards for personal or professional reasons. The ATI brand's image of exclusivity and quality plays a viutal role in the company's marketing strategy.
Having taken this into account, consider the Linux user community's reputation for using "hacked" or "modded" hardware for all sorts of reasons from saving money to illegally circumventing copyright restrictions. It follows that it is totally in ATI's interest to release drivers that work with their hardware exculsively. To do otherwise would be to associate the ATI brand with all matter of hacked, downscale, and jerry-rigged hardware, a move which would ultimately prove a detriment to ATI's profits.
If you go into a computer store and ask for an OEM ATI chances are you'll get a white box made by some unknown company that just took the ATIs chips and slapped them on the board , even though technicly OEM meens its made by ATI. this isn't a bad thig as 99% of the time "OEM" stuff works just as well as retail. this is more of a case of bad ATI drivers doing poor bios checking
I have an Excalibur (ATI 9000 Pro based card), and have to run through a series of installs
to stop the video card locking up my Win2K system. The original drivers seem to be buggy.
When it is working fine (like now), it is a damned fine graphics card, its just such a
bitch to get going.
http://jesus.everdense.com/
My answer: For business use, buy Matrox.
I was wondering how far down I'd have to scroll before I found another Matrox supporter. My G400 has run flawlessly on three different distros, and about 5 different XFree86 versions. It may be getting a bit long in the tooth, but I still haven't found a compelling reason to upgrade, and especially not to change GPU sets. I had an Nvidia system sitting alongside the Matrox box for several months, and both had identical monitors. The Matrox box was very easy on the eyes while the Nvidia always looked fuzzy. The Matrox white backgrounds were solid white while the Nvidia painted rainbows and shifting Moire patterns.
As to gaming, I'm playing X-Plane on the Matrox box using the latest Wine RPM. The frame rates are the same as they are on the Nvidia in Linux - I don't know how the two cards compare under Windows since I don't use Windows for anything else other than to supply Wine some DLLs. So with frame rates being equal, the Matrox wins for clarity of display, better drivers, and a more open philosophy. One possible drawback for some - Matrox's OpenGL drivers for Windows are not very good, but that doesn't affect me. YMMV and all that....
I have a "powered by ATI" card that I bought on the cheap, and I remember reading at rage3d.com that one particular release of the windows drivers for XP did a similar check for "OEM vs Retail". This was back in the day when the retail drivers and OEM drivers were different. Funnily enough though, my OEM card had a retail bios so I didn't experience this problem.
This problem in the older driver sets was removed (aain I cna't confirm) when ATI went to the unified driver "Catalyst series"
Maybe this set of drivers has been ported from the old code base? Now according to the press release the Linux build is a "unified driver". So I expect it is ported from the newer code base...
If you go to the ati site and click on the "powered by ATI" drivers, there is no option for a linux driver. It only appears under the "Built by ATI" drivers section. This would suggest to me that it is very deliberate. All of us can assume why... but none of us know for sure.
My gut feeling is they can't be sure how the OEM cards are set up (eg mem speed etc) and therefore can't guarantee the driver will work. ATI don't have the resources to field calls from every man and his dog world wide for 100 variants of the same card. Then again like I said this is only speculation. We should probably find out the reason before everyone shoots off at the hip and accuse ATI of all sorts of things.
Back to windows for a second. The solution to work around the windows install was a simple modify of an inf file....
Mybe it is just as simple for the linux xfree drivers, but I don't want to start pulling rpm's apart and looking at whats inside.
lounge around on the blue couch
My brother got a Gateway computer that was suppose to have an AGP Rage Pro II. For weeks it didn't work with X windows. One day he picked the old Mach64 X server by mistake and low and behold it worked. The OEM was selling what was essentially a Mach64 (it had more ram than a stock Mach64) and calling it a Rage Pro II (yes, I know the cards where probably simular, but there are sufficent differences to matter. Especially for 3-D games). You can imagine the problems this would cause. If your Wiz bang Radeon 9000 is really a first gen Rage 128 with 64 megs of DDR slapped on, but still reports itself as being a, so t Radeon The driver takes the card at it's word and nothing works.
As a side note, the "SoundBlaster Live!" that was suppose to be included has an Ensonique Audio PCI chipset instead of the EMU10K chips it was suppose to have. OEMs suck.
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
That argument makes zero sense.
1. Nobody, Linux hacker or otherwise, builds cards in their basement with modern surface mounted chips, it would cost dozens of times more than the card itself does.
2, They still have to buy the chips from ATI (if they have some other chips that ATI's drivers are useful for then ATI is in much worse trouble that from from your fantasy Linux hardware builder.)
3. The fix for the other cards was simple enough to be totally trivial for anybody capable of building the card from scratch.
You have to do better than this nonsense to try to insult Linux users. Go back to school.
3D graphics cards are really powerful computers
True, but they are not general purpose computers. They are designed to do one thing only - perform operations relevant to rendering 3D scenes. More than that, in fact - they are built to accelerate Direct3D and OpenGL operations specifically. Modifying the drivers might well allow you to do other cool things, but you'd almost certainly be better off doing those things with a normal CPU.
Other than that, while I sympathise with your sentiments, and to some extent agree with them, we don't really have much choice. The only fully working Linux drivers for modern graphics cards are closed source. By "fully working", I mean with complete, stable, fast suppot for all of the card's features. I'm pragmatic; if I've spent £200 on a new card, I want it to work properly. If that means using a close-source driver, so be it.
Finally, you seem not to realise that it isn't always up to the card/chipset manufacturers to open their driver source. NVidia, for example, is under NDA with several third parties over technology used in their cards and drivers. That means that they can't open the source to their drivers.
It's official. Most of you are morons.
here on the DRI web site is a little explianation about ATI naming
specificaly:
The difference between the 8500, 8500 LE, and the 8800 is clockspeed. The 8500 LE is made by third party manufacturers.
I have always used ATI cards, but at the moment I only have a ATI Radeon, so I can't try ATI's new drivers. I will probably upgrade for ut2003 and DeusEx 2. But any way I have a lot of confidence in the DRI people.
From: http://www.linuxfocus.org/English/January2002/arti cle222.shtml#222lfindex3
Currently most Linux graphic card drivers (X servers) do not support hardware-accelerated GLX/OpenGL for remote applications. They do support hardware acceleration for local applications. The effect is that remotely started OpenGL applications are hardly starting at all and are really slow. An exception are the closed source NVidia drivers. They have a direct rendering interface which supports indirect rendering for remote applications.
I use a central server to run my applications and then use X to display them remotely. Is the above excerpt out of date or do any other board manufacturers plan to incorporate the ability to run OpenGL apps from a server?
Brian.