Re: dri problems on radeon 7500 [M7] doing ufo:ai

Eric wrote:
> ...first I hope Im not offending anyone placing my question here. I
> as well as parts of the devel-crowd of ufo:ai (http://ufo.myexp.de)
> tried to track down the problem discribed below but failed to make a
> part of the game nor my config responsible. So as it looks like its
> more a problem with the oss driver than the config or game I thought
> this might be the right place to find assistance / hints.
Yes, you're probably right here.

> The System:
> 
> - -ibm t30 / radeon 7500 mobility [M7] - -debian etch / x.org / snv
> ufo:ai - -glxinfo / xorg.conf / glxgears look absolutly ok ( but if
> its any help I will send (relevant parts) of them cause)
> 
> The Problem:
> 
> ufo:ai is an oss game using the famouse quake 2 engine (in an
> extremly modified way through), rendering is done via glx.
> 
> on bootup everything, including the open_gl renderer, is started 
> perfectly  except those lines: [code] ...GL_EXT_point_parameters not
> found ...using GL_ARB_multitexture ...using
> GL_EXT_texture_env_combine ...GL_SGIS_multitexture not found 
> ...ignoring GL_ARB_texture_compression ...using
> GL_EXT_texture_filter_anisotropic [16 max] [ 1 selected] ...using
> GL_EXT_texture_lod_bias ...using GL_EXT_stencil_wrap 
> ...GL_EXT_fog_coord not found ...GL_ARB_fragment_program not found 
> ...GL_ATI_separate_stencil not found ...max texture size: 
> ......detected 512 ...SDL_ttf inited [/code]
There isn't really a "problem" here, just reported what the chip/driver
can do.
Now, I don't know what the game actually requires, but I'd hope it's not
going to depend on ARB_fragment_program or ATI_separate_stencil, the
chip simply can't do that. It shouldn't need GL_SGIS_multitexture
neither, this extension is deprecated and afaik completely replaced by
ARB_multitexture (I don't think it ever was an official extension?).
If you upgrade to a new dri driver, it will support GL_EXT_fog_coord.
The chip can't do EXT_point_parameters (it can't even do large points),
the app should detect that (and it seems it does) and use "standard"
opengl primitives instead (at a performance hit obviously).

> Ingame itself problems vary from missing texture, wired fonts to mess
> up models leading to a completly unusable programm.
Not sure about the fonts problem, but the missing texture problem sounds 
like it could be related to the max texture size announced only
being 512. This would be an app error (i.e. while it would detect max 
texture size is 512 it still would be trying to use larger textures - 
mesa would reject such requests completely). However, a newer driver 
version (I'd suggest mesa 6.5.1 rc) should also help there, as you can 
force it to announce support for larger textures (which might cause an 
incredibly slow raster fallback, but only if you're unlucky). Also, 
starting your desktop in a lower resolution might help, as more ram will 
be available for textures and the announced max size may be larger. I 
think increasing GARTsize in your xorg conf might also help, but I'm not 
sure there.

> People using the nvidia/ati probitarian drivers after encountering
> those problems fixed em doing so. which obvious isnt an option to me.
> 
> 
> If anyone has a hint how to improve the situation or what might be
> the reason for the problems discribed, if they are driver-related (as
> the dev department of ufo:ai said) I would be vary grateful. And so
> will others as meanwhile it turned out other people with same
> hardware / driver config have exactly same problems. Eventualy,
> thanks for your time and hope I didnt ignore any list rules. Best
Depending on what the "messed up models" problem is, you could try 
tcl_mode=0 (or maybe tcl_mode=1), but in general I'd suggest you just 
try a newer dri driver.

Roland

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
--