pouët.net

ATI Catalyst OpenGL error...

category: general [glöplog]
I installed both original ATI and Omega drivers (both Catalyst 7.9.1 based) on my Windows XP Professional SP2 machine. A very strange problem occurred when I programmed a graphic layer which abstracted the OpenGL and Direct3D API: MY OpenGL windows were all blank, as if SwapBuffers() never did anything.
I've programmed with both OpenGL and Direct3D before, but this never happened.
So I checked with gDEBugger, and it found no error at all. Later, after several hours of hacking and debugging I decided that my Windows Installation was fucked up, and reinstalled XP. Nothing changed when I finally had installed everything from the ground up.
Next thing was googling the error, but I didn't find anything useful, except someone saying that he had the very same behaviour in windowed, but not in fullscreen mode; and that this was NOT happening with his older Catalyst 7.4 drivers. So I switched back to 7.4, and I could see all my OpenGL programs drawing to the window again.

Now I wonder: Are there not enough people complaining about this problem? Why is it so hard to find information regarding this bug? Does ATI know that their newer drivers fuck up OpenGL in windowed mode (found nothing on their website)? Is nobody using OpenGL anymore? Or is my system configuration such an exotic one that no one else experiences those things? (Intel 975X Chipset, Core2Duo E6600, 3 Gig RAM DDR2-800, ATI X1900 XT...)

To put the long story short: Have any of you had the same problem?
added on the 2007-10-25 02:24:09 by hcdlt hcdlt
Forgot to mention: OpenGL did not draw into the window, but Direct3D did so flawlessly with the same driver version.
added on the 2007-10-25 02:25:42 by hcdlt hcdlt
hcdlt: Yes but I eventually had enough with lamers at ATI not being able to support OpenGL properly over the years, which led me to switch to DirectX and to never buy an ATI card.

About your problem, I gave up trying to fight crappy ATI drivers long ago, considering it a waste of time, good luck though :)

I guess coders still using OpenGL might be able to help you (Iq/RGBA, Navis/ASD, Kurli/Kewlers or Ithaqua/Stravaganza)
added on the 2007-10-25 02:31:54 by keops keops
You seem to jump to conclusions, perhaps you simply have a bug (or most likely twenty) in your gl-setup code that you just didn't catch yet?
added on the 2007-10-25 02:32:44 by kusma kusma
Kusma: it's not like it's new that ATI cards still have a shitty OpenGL support in 2007 :)

Plus, if I were the only one pissed off with that I would have tried to lose even more time trying to figure out what was wrong but after talking with a few other coders at work and on the scene, I found it wiser to forget about the duo ATI-OpenGL.

(my problems were not the same as hcdlt's, but they were equally annoying, disappeared and reappearing with time, depending on new drivers versions, not always in the good direction unfortunately).

Strangely enough, I never had those problems with NVidia cards.
added on the 2007-10-25 02:43:58 by keops keops
keops: Dude, he has issues getting swapbuffers to work.
added on the 2007-10-25 02:45:27 by kusma kusma
question is - do any other GL applications work?
added on the 2007-10-25 03:46:38 by Gargaj Gargaj
kusma: I have to admit I did not even read his post entirely :))

I guess it kind of shows how pissed I growed with ATI-OpenGL issues over time.
added on the 2007-10-25 03:46:43 by keops keops
no problems with opengl windowed using catalyst 7.7 (=definitely not the newest release). some code would help finding the problem - just because it doesn't occur with older drivers doesn't mean your code can't be wrong.
added on the 2007-10-25 03:52:15 by ryg ryg
I discovered not long ago(by reading the blue book) that
glScissor, glColorMask, glDepthMask, glStencilMask affect the glClear command.
Maybe it's one of those, but that would mean the GL implementation is not setting the right default values.
added on the 2007-10-25 08:25:55 by duffman duffman
Quote:
question is - do any other GL applications work?


Obviously, yes, so the problem is in his application, not in the drivers...
added on the 2007-10-25 09:26:24 by nystep nystep
he never stated that explicitly (the post is a bit ambiguous)
added on the 2007-10-25 09:49:14 by Gargaj Gargaj
I know 7.9 Cats(maybe 7.9.1 Omega fixed the issue though?) did some real funky crap on my system although for me DX stuff wouldn't work but i think OGL did heh, so reverted back to 7.8 and all works fine now. Doubt that is the issue you are having but /shrug
added on the 2007-10-25 10:40:10 by Intrinsic Intrinsic
Haven't seen or heard of that problem before. (And not here on cat 7.8/9/10/11b, x1900). Tried drawing to the front buffer?, without double buffer, disabling z/whatever tests etc?

Generally I'd say that at least the x1x00 drivers are very stable atm, the hd2x00 series is another story, at least for the shader compiler.



added on the 2007-10-25 10:45:19 by Psycho Psycho
You should basically stay the fuck away from Omega-drivers, and father use normal catalyst (use Mobility Modder to get it to work on laptops) instead. Omega-drivers has been known to "tweak" driver-parameters (or whatever they do) that prevents vital stuff to conform to the spec.
added on the 2007-10-25 10:48:01 by kusma kusma
Omega drivers ...
stay away from kids raping your registry settings ...
and i never believed registry/settings/whateveryoucallthem hacks could reliably improve the performances ...
Yeah i stay away from Omega now too, I just use the driver with ATI Tray Tools, avoid that CCC bloated junk.
added on the 2007-10-25 11:07:38 by Intrinsic Intrinsic
No idea if mobility modder does anything really useful, but to install the standard ati drivers on a mobile chipset, just run the installer then cancel it when it loads up. Then use device manager to update the drivers manually, and point it at the nice driver folder the ATI installer left. Works perfectly (here anyway), and makes you wonder why ATI are putting an apparently pointless bit of code in the installer to stop you using it!
added on the 2007-10-25 11:43:38 by psonice psonice
psonice: it just fixes the installer so you can run it without hacking it yourself. I guess the reason why ATI is preventing laptops from installing the drivers is support; if they let you install it, they must provide support for it. If they don't, you'll have to get support from your laptop vendor instead.
added on the 2007-10-25 11:48:34 by kusma kusma
hmm i don't know if it's only a matter of support
i tried to installed the latest nvidia drivers on a mac book pro under vista and the nvidia detection tool says i need to go to the vendor sites because they might have modified the driver to make suspend, and stuff like power saving work
CCC is total crap! seriously. eats up >50megs of RAM and CPU cycles for NOTHING. But the new Nvidia control-panel stuff sucks bigtime too. Its supposed to be for total noobs, seriously oversimplified and therefore not useful anymore.
And whats so wrong about making it accessible the Display settings dialog? I don't understand those people...
added on the 2007-10-25 14:07:16 by raer raer
just to make things a little bit clearer:

no, not just my own OpenGL applications didn't work anymore, it was EVERY SINGLE OpenGL app i tried (plastic demos, ASD demos, GL Excess, the complete neHe tutorial list and so on), when started in windowed mode. so I didn't just jump to conclusions, believe me. (I spent several hours in Application and OpenGL debug before!)

the omega driver thing was just a desperate attempt to solve the issue.

But seeing some of you using a similar system configuration with newer Catalyst drivers, I figure it might be something wrong with my hardware. (a very strange thing though, OpenGL not drawing while Direct3D working perfectly...)
added on the 2007-10-25 16:22:37 by hcdlt hcdlt
@psycholns:

when trying to draw to the front buffer, without double buffer or z-buffer en- or disabled (debug said no to all of these), I should at least have seen a somewhat colorful or maybe completely black picture drawn to my window (read: uninitialized front or back buffer). but that wasn't the case. The only thing I saw was the default window background (or the desktop background if the window was transparent).
added on the 2007-10-25 16:27:08 by hcdlt hcdlt
so let me get this straight -- for one platform you support 2 graphics api's
why oh why

(why anyone would use ogl when there's d3d avail scares me as well)
added on the 2007-10-26 15:33:13 by superplek superplek
performance would be an obvious reason (esp. batching performance)
added on the 2007-10-26 16:32:00 by Psycho Psycho

login