FGLRX Pixel Buffer support

[sandain]

I'm not following here. If WoW uses this extension, but the driver doesn't implement it, how on earth does it render correctly in Windows in the first place?

Because the Windows driver most likely implements this feature, or because very few Windows users actually run the OpenGL client.

[dungeon]

Oh, oh... Same thing i experianced when ATI droped support for pbuffers for my 9250 (2005 with 8.15 driver), with argument "it is hard to maintaining any further"... One year later my card was not supported anymore. The end. NVIDIA for compete and many years older chips have support even today! So, i just don't know when ATI think to stop
playing like that with their users?

@sandain
Here are some 'fgl_glxgears' screenshots from that time, with or w/o pbuffers driver support. I remember if that is good, wine use that and translate as wgl pbuffer. You can test this also if you think that driver lack pbuffers.

[sandain]

After playing around with the Wine code for a while, and getting no results, I decided to check how things performed in Windows. I am now back to believing this to be a bug on AMD's end. This bug also shows up in Windows when using the OpenGL client with WoW. I tested this with versions 8-6 and 9-1, with the same result.

Windows - OpenGL:


Wine - OpenGL


Wine - Direct3D


Note: Wine renders the Direct3d client the same as Windows. The OpenGL client works with Nvidia drivers on either platform.


Update: This bug has been fixed as of FGLRX 9-8. Thanks AMD!

[dungeon]

Yes, this is fglrx bug. Games often uses pbuffers for map viewer, you can see that in Dungeon Siege 2 also.

P.S. But then, i think you have translated WGL_ARB_pbuffers (you can use 'glview' tool, to see if Wine exported that extension), because WoW DX client works good. So, if problem is only with opengl stack of wine's pbuffers, bug can be in both, but as you stated that opengl client
run the same in windows, i think there is no hope with fglrx.

[Qaridarium]

WINE: "fixme:d3d_shaderrint_glsl_info_log Error received from GLSL shader #16: "Fragment shader was successfully compiled to run on hardware.\nWARNING: 0:2: extension 'GL_ARB_draw_buffers' is not supporte"
"

i testet it on OFP in Wine and Aquamark3benchmark on the catalyst 8.592 and same problem on 8.582.

[bridgman]

Might be a bogus message, not sure :

http://lists.apple.com/archives/Mac-.../msg00048.html

[Qaridarium]

Might be a bogus message, not sure :

http://lists.apple.com/archives/Mac-.../msg00048.html

for me OFP (a nice game) works most of the time..
i think there only was some litle deteils that was not renderet..
i can bring some fotos online winxp VS Linux...
there are some diverences..
i think on nvidia there are nerly no diverences.. ..
but i can't test this becourse all my nvidia carts are death..
a frend send me Pictures of OFP on windows he has a nvidia...
and yes there are diverences... but it can be the "wine" what makes the diverence not the driver..
so over all.. i don't know..

[guruman]

I know this is a somewhat old thread but i am experiencing the same problem as the original author (wow minimap all white indoors).

Now im using the 9.5 drivers and GLX_SGIX_pbuffer is exported both in the server and client side of GLX, fgl_glxgears works with no problem, so my question is basically is this then implemented now?

And if so then i suppose my problem is the WGL -> GLX wrapper in wine or perhaps the detection of the extension by wine?

Wine version is: 1.1.22 running on ubuntu 9.04

[guruman]

To answer my own question:
Yes this seems to be implemented now.

No the problem is not with Wine, its with ATI's driver.
It's supposed to include the extensions that are present in the driver in the "GLX extensions:" list which it doesn't.

Why the driver doesn't do this is another question which i don't have the answer to, seems stupid since its there both at client and server side.

/BR

[bridgman]

Can you be a bit more precise about what is not working with the extension ? Maybe sample code or something ? I'm not clear how the extension can be "not implemented in the driver" if fgl_glxgears is working.

My understanding from the devs is that the extension was always implemented in the driver code (and worked in other OSes) but since it was only a standard extension in Windows the X binding was not added until recently.