UFO:Alien Invasion
Archive => Bugs prior to release 2.4 => Topic started by: Edi on May 26, 2010, 09:21:15 pm
-
The black models issue on ATI cards is still not squashed, despite being reported as fixed.
I compiled revision 30167 today and I'm still getting black soldiers, aliens, crashed ufo fighters, civilians etc.
Windows XP Pro SP3 with ATI Radeon HD4870 and latest drivers.
Console log of video initilaization:
2010/05/26 22:24:58 ----- network initialization -------
2010/05/26 22:24:58 libcurl/7.16.4 zlib/1.2.2 initialized.
2010/05/26 22:24:58 ------ server initialization -------
2010/05/26 22:24:58 added 7 maps to the mapcycle
2010/05/26 22:24:58 ----- console initialization -------
2010/05/26 22:24:58 Console initialized.
2010/05/26 22:24:58 ------- video initialization -------
2010/05/26 22:24:58 SDL version: 1.2.14
2010/05/26 22:24:58 I: desktop depth: 32bpp
2010/05/26 22:24:58 I: video memory: 0
2010/05/26 22:24:58 I: Available resolutions: 1920x1200 1920x1080 1680x1050 1600x1200 1440x900 1400x1050 1360x1024 1360x768 1280x1024 1280x960 1280x800 1280x768 1280x720 1152x864 1024x768 848x480 800x600 720x576 720x480 640x480 (20)
2010/05/26 22:24:58 I: video driver: windib
2010/05/26 22:24:58 I: setting mode 0
2010/05/26 22:24:58 I: set swap control to 0
2010/05/26 22:24:58 I: 1920x1200 (fullscreen: yes)
2010/05/26 22:24:58 I: got 8 bits of stencil
2010/05/26 22:24:58 I: got 24 bits of depth buffer
2010/05/26 22:24:58 I: got double buffer
2010/05/26 22:24:58 I: got 8 bits for red
2010/05/26 22:24:58 I: got 8 bits for green
2010/05/26 22:24:58 I: got 8 bits for blue
2010/05/26 22:24:58 I: got 8 bits for alpha
2010/05/26 22:24:58 I: got 0 multisample buffers
GL_VENDOR: ATI Technologies Inc.2010/05/26 22:24:58 GL_RENDERER: ATI Radeon HD 4800 Series2010/05/26 22:24:58 GL_VERSION: 3.2.9756 Compatibility Profile Context2010/05/26 22:24:58 GL_EXTENSIONS: GL_AMDX_name_gen_delete GL_AMDX_vertex_shader_tessellator GL_AMD_conservative_depth GL_AMD_draw_buffers_blend GL_AMD_performance_monitor GL_AMD_seamless_cubemap_per_texture GL_AMD_shader_stencil_export GL_AMD_texture_cube_map_array GL_AMD_texture_texture4 GL_AMD_vertex_shader_tessellator GL_ARB_color_buffer_float GL_ARB_copy_buffer GL_ARB_depth_buffer_float GL_ARB_depth_clamp GL_ARB_depth_texture GL_ARB_draw_buffers GL_ARB_draw_buffers_blend GL_ARB_draw_elements_base_vertex GL_ARB_draw_instanced GL_ARB_fragment_coord_conventions GL_ARB_fragment_program GL_ARB_fragment_program_shadow GL_ARB_fragment_shader GL_ARB_framebuffer_object GL_ARB_framebuffer_sRGB GL_ARB_geometry_shader4 GL_ARB_half_float_pixel GL_ARB_half_float_vertex GL_ARB_imaging GL_ARB_instanced_arrays GL_ARB_map_buffer_range GL_ARB_multisample GL_ARB_multitexture GL_ARB_occlusion_query GL_ARB_pixel_buffer_object GL_ARB_point_parameters GL_ARB_point_sprite GL_ARB_provoking_vertex GL_ARB_seamless_cube_map GL_ARB_shader_objects GL_ARB_shader_texture_lod GL_ARB_shading_language_100 GL_ARB_shadow GL_ARB_shadow_ambient GL_ARB_sync GL_ARB_texture_border_clamp GL_ARB_texture_buffer_object GL_ARB_texture_compression GL_ARB_texture_compression_rgtc GL_ARB_texture_cube_map GL_ARB_texture_cube_map_array GL_ARB_texture_env_add GL_ARB_texture_env_combine GL_ARB_texture_env_crossbar GL_ARB_texture_env_dot3 GL_ARB_texture_float GL_ARB_texture_mirrored_repeat GL_ARB_texture_multisample GL_ARB_texture_non_power_of_two GL_ARB_texture_query_lod GL_ARB_texture_rectangle GL_ARB_texture_rg GL_ARB_texture_snorm GL_ARB_transpose_matrix GL_ARB_uniform_buffer_object GL_ARB_vertex_array_bgra GL_ARB_vertex_array_object GL_ARB_vertex_buffer_object GL_ARB_vertex_program GL_ARB_vertex_shader GL_ARB_window_pos GL_ATI_draw_buffers GL_ATI_envmap_bumpmap GL_ATI_fragment_shader GL_ATI_meminfo GL_ATI_separate_stencil GL_ATI_texture_compression_3dc GL_ATI_texture_env_combine3 GL_ATI_texture_float GL_ATI_texture_mirror_once GL_EXT_abgr GL_EXT_bgra GL_EXT_bindable_uniform GL_EXT_blend_color GL_EXT_blend_equation_separate GL_EXT_blend_func_separate GL_EXT_blend_minmax GL_EXT_blend_subtract GL_EXT_compiled_vertex_array GL_EXT_copy_buffer GL_EXT_copy_texture GL_EXT_draw_buffers2 GL_EXT_draw_instanced GL_EXT_draw_range_elements GL_EXT_fog_coord GL_EXT_framebuffer_blit GL_EXT_framebuffer_multisample GL_EXT_framebuffer_object GL_EXT_framebuffer_sRGB GL_EXT_geometry_shader4 GL_EXT_gpu_program_parameters GL_EXT_gpu_shader4 GL_EXT_histogram GL_EXT_multi_draw_arrays GL_EXT_packed_depth_stencil GL_EXT_packed_float GL_EXT_packed_pixels GL_EXT_pixel_buffer_object GL_EXT_point_parameters GL_EXT_provoking_vertex GL_EXT_rescale_normal GL_EXT_secondary_color GL_EXT_separate_specular_color GL_EXT_shadow_funcs GL_EXT_stencil_wrap GL_EXT_subtexture GL_EXT_texgen_reflection GL_EXT_texture3D GL_EXT_texture_array GL_EXT_texture_buffer_object GL_EXT_texture_buffer_object_rgb32 GL_EXT_texture_compression_latc GL_EXT_texture_compression_rgtc GL_EXT_texture_compression_s3tc GL_EXT_texture_cube_map GL_EXT_texture_edge_clamp GL_EXT_texture_env_add GL_EXT_texture_env_combine GL_EXT_texture_env_dot3 GL_EXT_texture_filter_anisotropic GL_EXT_texture_integer GL_EXT_texture_lod GL_EXT_texture_lod_bias GL_EXT_texture_mirror_clamp GL_EXT_texture_object GL_EXT_texture_rectangle GL_EXT_texture_sRGB GL_EXT_texture_shared_exponent GL_EXT_texture_snorm GL_EXT_texture_swizzle GL_EXT_timer_query GL_EXT_transform_feedback GL_EXT_vertex_array GL_EXT_vertex_array_bgra GL_IBM_texture_mirrored_repeat GL_KTX_buffer_region GL_NV_blend_square GL_NV_conditional_render GL_NV_copy_depth_to_color GL_NV_explicit_multisample GL_NV_primitive_restart GL_NV_texgen_reflection GL_SGIS_generate_mipmap GL_SGIS_texture_edge_clamp GL_SGIS_texture_lod GL_SUN_multi_draw_arrays GL_WIN_swap_hint WGL_EXT_swap_control2010/05/26 22:24:58 2010/05/26 22:24:58 using GL_ARB_texture_non_power_of_two
2010/05/26 22:24:58 using GL_ARB_shading_language_100
2010/05/26 22:24:58 GLSL Version: 1.50
2010/05/26 22:24:58 using GL_ARB_framebuffer_object
2010/05/26 22:24:58 max draw buffers: 8
2010/05/26 22:24:58 max render buffer size: 8192
2010/05/26 22:24:58 max color attachments: 8
2010/05/26 22:24:58 using GL_ARB_draw_buffers
2010/05/26 22:24:58 max supported lights: 8
2010/05/26 22:24:58 max texture units: 8
2010/05/26 22:24:58 max texture coords: 16
2010/05/26 22:24:58 max vertex attributes: 16
2010/05/26 22:24:58 max texture size: detected 8192
2010/05/26 22:24:58 Warning: high resolution globe textures requested, but could not be found; falling back to medium resolution globe textures.
2010/05/26 22:24:58 R_LoadProgram: 'world' loaded.
2010/05/26 22:24:58 R_LoadProgram: 'warp' loaded.
2010/05/26 22:24:58 R_LoadProgram: 'geoscape' loaded.
2010/05/26 22:24:58 R_LoadProgram: 'combine2' loaded.
2010/05/26 22:24:58 R_LoadProgram: 'convolve3' loaded.
2010/05/26 22:24:58 R_LoadProgram: 'atmosphere' loaded.
2010/05/26 22:24:58 R_LoadProgram: 'simple_glow' loaded.
2010/05/26 22:24:58 SDL_image version 1.2.6
2010/05/26 22:24:59 SDL_ttf version 2.0.9 - we need at least 2.0.7
One thing did occur to me: Does the fix require a clean compile of all maps to work correctly? I figured that might be the case, if it is related to the lighting issues, as it seems to be from some of the code commentary.
-
Compiled 30169 and seeing this still myself.
Black models only when realtime lighting is enabled.
Shaders enabled without realtime lighting is fine - however this then leads to a transparent Earth when you get back to the Geoscape.
edit: Using Windows 7 32-bit on AMD PhenomII X3 720 with an ATI 4850 gfx card using Catalyst 10.3 drivers (OK, should probably update them...)
-
please retry with latest rev.
-
Haven't been able to retry with latest build, largely do to not being able to load into any maps. Most the time it just *blinks* back to the geoscape without anything happening. Or it loads the map then essentially freezes, causing me to have to alt-F4, or use task manager to end task..... but a caveat to that is.... I haven't ever been able to successfully enter a tactical map on my desktop... not quite sure why(usually freezes, today first time did the blink thing). But my laptop can enter tactical. However, I think last build I was using on my laptop was something like 30105 or something like that. But on my laptop, before the attempted patch of rev 30155(I do use ATI on both desktop(4670) and laptop(5870),) depending on zoom level, soldier/alien/ufo textures were either black(zoomed out) or a kind of shimmery, almost TRON like blue(when zoomed in).
Anyway, tomorrow I ought to be able to test a build or two to give more constructive feedback.
-
Just tried with 30175 - no luck, still the same problem.
edit: Also updated graphics card drivers, still no luck.
-
As of r30175, I don't have black models here.
Mine are glowing. See screeny.
Been like that for some 100 revs.
-
duke what is your hardware (graphiccard)?
-
As of r30175, I don't have black models here.
Mine are glowing. See screeny.
Been like that for some 100 revs.
That's what they looked like on mine, but I think they were blue, though I may be mistaken.
The only hardware i've successfully gotten into a tactical map was on a mobility 5870
-
As of r30175, I don't have black models here.
Mine are glowing. See screeny.
Been like that for some 100 revs.
hehe, your soldiers are burning! :D
-geever
-
Finally figured out how to get into a tactical map without it freezing. Anyway, my models are a bit wonky, but all I have to do is disable post-processing to fix. I am able to leave the GLSL box ticked.
-
Compiled 30169 and seeing this still myself.
Black models only when realtime lighting is enabled.
Shaders enabled without realtime lighting is fine - however this then leads to a transparent Earth when you get back to the Geoscape.
edit: Using Windows 7 32-bit on AMD PhenomII X3 720 with an ATI 4850 gfx card using Catalyst 10.3 drivers (OK, should probably update them...)
Confirmed on Windows XP Pro SP3 Radeon HD4870 (Catalyst Control Center 10.4), UFOAI revision 30181.
With GLSL shaders (and postprocessing) enabled AND realtime lighting enabled, black models on battlescape.
When GLSL shaders OR realtime lighting is turned off, models become normally visible
When realtime lighting is turned off, geoscape is transparent
Thus, it is possible to use realtime lighting as the workaround, because turning it off and on does not require a restart but GLSL shaders do or geoscape gets screwed up.
-
try r30174 and above
-
try r30174 and above
I've currently got revision 30181 installed and it's behaving as described in my previous post. Having both shaders and realtime lighting on results in black models.
-
please add this info to the bug tracker item.
-
duke what is your hardware (graphiccard)?
ATI Radeon HD 4650.
Realtime lighting, GLSL shaders and post-processing are the only boxes ticked in the video options.
-
please add this info to the bug tracker item.
Done. I will add the console log excerpt from my latest install later.
-
Would it help if I figured out the revision that broke it ?
-
It could, as knowing when something broke is often as important as knowing what it was that broke. If somebody can add how it broke as well, then finding a fix may become possible.
-
Uhm...Thx Edi...I already knew that ;)
The implicit question in my post was: Do Mattn / arisian already know what it broke (and are just figuring out how to fix it) ?
btw in r30285 the glowing persists, but the color has changed to a nice blue...
-
Man already playtesting 285? Committed only 7minutes ago :D
First time I noticed the shader issue on my laptop my stuff showed up blue just like that. Now for some reason it's mainly just black, with a bit of shimmer like the blue around some edges and whatnot.
-
Man already playtesting 285? Committed only 7minutes ago :D
There is no magic in that coz it was *me* who committed 285 ;)
-
Uhm...Thx Edi...I already knew that ;)
The implicit question in my post was: Do Mattn / arisian already know what it broke (and are just figuring out how to fix it) ?
btw in r30285 the glowing persists, but the color has changed to a nice blue...
Uh, sorry about that, I tend to get a little literal sometimes.
Nice team of Mr. Freezes you have there... :D
-
Uh, sorry about that, ...
Nevermind, it just made me smile ;)
btw you're doing a great job as a '1st level support guy' / 'poweruser' on these forums. Keep it up :)
-
Nevermind, it just made me smile ;)
btw you're doing a great job as a '1st level support guy' / 'poweruser' on these forums. Keep it up :)
Thanks. I try. Knowledge is of no use unless it's shared.
Having a knack for documenting things and putting them in readable format helps a lot. Bending it out of iron bars, as it were, or that's the rough translation from Finnish. :)
-
i run ufoai on ubuntu/lucid with ati's driver.
#lspci
01:00.0 VGA compatible controller: ATI Technologies Inc M92 [Mobility Radeon HD 4500 Series]
i'm at rev 30293, and i get the black things when i activate realtime lighting.
-
According to changelogs for 380 and 381, the shader issue was fixed for ATi. It has not, atleast for me. I built 2.3 rev 30381 and my battlescape models are still a shimmery red. Once I disable Post-Processing, it works fine.
screenie attached.
-
Yeah, I'm still trying to figure out what's causing this; once I do that, I'm fairly confident I can fix it. The problem is really that it seems very system-specific; I can't seem to replicate these "glow" issues on a ATI 4670 card (Windows XP, driver version 8.723). This makes it very difficult to debug, since I can't tell what's wrong, and I can't tell if I've fixed it ???
It might be helpful to me if people who are experiencing this issue could post the following:
1) Your system specs, including CPU, graphics card, driver version, operating system, and anything else you think might be relevant
2) A screenshot of the problem on your system, since it seems to look different for different people
3) A copy of your logfile
4) What conditions the problem does/doesn't happen under (e.g. does postprocessing on/off make a difference, or just realtime lighting? Does it only happen under certain gameplay conditions? Is it consistent, or does it seem to occur at random?)
Also, be sure that you're working from the most recent version of the 2.3 branch (since my first priority is getting that stable, that's where the changes will happen first).
I'll try to get this sorted out as quickly as possible; I appreciate everyone's patience with this.
-
Core 2duo 5200
ATI 4650(ithink, might be 4670) driver 8.682.2.0 (It's been a while(probably before rolled over to 30000) since I was playing on my laptop, which is a mobility 5870, but I seem to recall it being the same issue with shiny dudes)
I'll have to get you a logfile later, not sure if my recent one would have my battlescape.
Screenie was earlier, back before the original attempted ATi hack, my people were blue, now they red.
I have all options on pretty much, all I have to do to get it to fix is turn off postprocessing. However, the Black models IS fixed. Before I had to turn of GLSL also to fix black models.
I haven't actually tried multiple battlescape with options on to see if it's only certain maps, I'll do that over the next few days. And since most of the commits to 2.4 recently strike me as WIP, I'll stick to 2.3 for a bit.
EDIT: After perusing changelogs, I see the render_work branch. How would I go about integrating that into my code to compile, or is it even ready for primetime testing?
-
I expect to see whether the problem will have disappeared for me once I compile 30397 (now in progress).
System specs:
Graphics
ATI Radeon HD4870 with Catalyst CC 10.5
Driver version 6.14.10.7082 (from msinfo)
Driver Packaging Version 8.732-100504a-099992C-ATI (from Catalyst CC)
OpenGL Version 6.14.10.9836
Other
Windows XP Pro SP3
Core 2 Duo E6600 processor
Asus P5B-Deluxe motherboard
4 GB RAM
-
Core 2 Duo 5200
Win 7 64bit
Radeon 4600 series(cant recall 4650 or 4670) driver 8.682.2.0 (Dxdiag)
MOBO Gigabyte g31m
Disabling EITHER real-time lighting OR post-processing fixes my colorful guys, which turns out the colors change sometimes(pic below)
Side note, for some reason my graphics selector for maximum lights is stuck on 1. I can change it all I want, but goes back to 1.
CRAP, forgot to post a console, and I've restarted to fiddle with graphics some more and have run out of time. In a couple hours
REV 30394 btw
-
Nice. We didn't have green guys yet, did we ?
-
So it's not just my models, or just alien models, but it's not everything out there. SS below.
And here's my console, started game with GLSL, realtime on, and post processing on, then i turned off post process. Then post back on and realtime off. Then back to all on.
-
At revision 30427 I'm still seeing no change whatsoever.
GLSL Shaders ON
Postprocessing ON
Realtime lighting ON
Result: Black models
*******
GLSL Shaders ON
Postprocessing ON
Realtime lighting OFF
Result: Normal models
*******
GLSL Shaders OFF
Realtime Lighting ON
Result: Normal models
Just as it was before. My system specs are posted earlier in this thread.
ufoconsole.log attached.
-
For those with black models, how many dynamic lights are active? (it's a counter near the bottom of the graphics menu)
-
Well I don't have black models, I have the shiny guys. But My lights when on a 2.3 build seems to be stuck at 1. I can turn it up all i want, but drops back down to one.
-
For those with black models, how many dynamic lights are active? (it's a counter near the bottom of the graphics menu)
For me it's one. Haven't tried to change it.
-
I know that this is the 2.4 trunk forum, not 2.3 branch. But the recent change by arisian rev 30502.... I have no issues with glowing people or black people or anything else. All options on; GLSL, real time lighting, post-processing, 8 dynamic lights etc.
:D
Now question I kind of have.... what exactly did he remove? Meaning, was is like performance improving style things? Completely new kinds of shiny eye-candy? Just curious, you know?
-
It's working now? Hooray! To answer your question, I changed a couple of things. First, the only major "feature" I removed was the directional illumination of actors and other non-tile based entities. They now look the way they used to in terms of lighting. The glow maps are still used.
In terms of what fixed the issue, I'm still not really sure which of the changes I made actually did it. I changed how glow was handled by multi-stage materials, but that only applies to brushes (it should fix a bug where glow would shine "through" other objects). I also made several small changes to the code that is responsible actually rendering the glow effect in the hopes that one of them would fix the issue with the randomly glowing models. I'm just glad that it seems to have done the trick; I really wanted to get this fixed before the 2.3 release.
If anyone else is still having issues with this (in the most recent 2.3 branch), please let me know as soon as possible so I can try to iron all the bugs out before the release.
-
With r30504 in 2.4 branch I still have the effect if I turn on post-processing. Is that according to plan ?
-
Probably, since 2.4 is now the development, while 2.3 is intended for public release soon, so they have to get as much as humanly possible working before the can release. The arisian commit was only for the 2.3 branch.
-
Yes, I haven't done any work on 2.4, I'm concentrating on getting the 2.3 codebase patched for release. I'll try to move the changes into 2.4 when I get a chance.
-
Fine with me.
-
to find out which of the changes fixed the problem we should merge them in small steps into trunk. make more smaller commits to get closer to the origin of the problem.
-
Agreed, I just wanted to get it working any way possible for 2.3
-
Which is good, I've been playing most all today with 30502. No issues at all I've particularly ran across. I am curious though what exactly you "disabled" kind of thing? Was it things to make game more pretty? Or things to make run faster? (I ask the faster question because my framerate is quite a bit lower(15-20 on average in battlescape) than I experienced before, though I now am able to play with all the pretties on). So just curious. But it DOES work! One of these days I'll remember to take a screenie of the forest map, the water textures are a bit wonky looking when scrolling.
-
There were a few "optimizations" that I took out in places where I think they may have been skipping some cases that should have been handled. Making them handle everything was the easiest way to fix it for the release, even if it's not the most optimal way to do so.
Alternatively, if you've never actually used postprocessing before, that would explain it, since postprocessing in the current implementation is known to be very expensive in terms of framerate. I'm not quite sure why it's so slow, to be honest; it's something I'll try to fix if I can ever figure out what's causing it.
-
I actually can't quite recall if I was doing post-processing off, or if it was realtime-lighting i ticked off to fix the shiny model bug on ATi. Another potentiality is my computer and graphics card aren't exactly top of the line... I built this for about 500$ a year and a half ago. Core 2 Duo 5200 (2.5ghz) and a Radeon 4650. In a day or two I'll get around to installing on my laptop. Much higher specs :D Core i7, Mobility 5870 etc.
Actually might be able to run faster if it wasn't running in windowed mode, even if said resolution isn't screen max. I multitask a lot, and my mouse basically won't release once I've alt-tabbed.
Anyway, aside from the slowness(which I can manage with), the game looks awesome to me. And definitely brings back the good ol X-Com memories!
-
Was this problem supposed to be fixed in trunk? I'm still experiencing it with the öatest 2.4 dev builds, no change.
-
No, it should be fixed in the 2.3 branch, but I haven't fixed it in trunk yet.
-
No, it should be fixed in the 2.3 branch, but I haven't fixed it in trunk yet.
Thanks for the informative comment on the bug tracker. Much appreciated. :)
My personal workaround for this is to map F5 and F6 to turn real time lighting off and on at a stroke. I like the GLSL eyecandy on the geoscape far too much to part with it and real time lighting does not require a restart when turned back on. Enter battlescape, press F5 and ta-daa, no more black models. Come back to geoscape, press F6 and you have eyecandy back. The required code for keys.cfg is
bind F5 "set r_lights 0;"
bind F6 "set r_lights 1;"
-
Still no realtime lighting on ATI-gfx-cards, although my ATI-card's driver claims it is fully OpenGL 4.0 compatible :'(
Are there any news on that front ?
I saw you made a big update for the renderer_work branch, Arisian ! 8)
I wanted to try this branch, but unfortunately I was not able to compile any of the verions with codeblocks yet (errors out), but I was also wondering if maybe there are some fixes already made in that branch, which you could merge into trunk ?!?
Is there a way to get some screenshots of the renderer_work branch @ least ?! Or is it a secret until finished ? Give us some updates !!! ;)
Greetinx.
-
The work I'm doing in the branch is a pretty major overhaul; there's no easy way to integrate just part of it into trunk. I don't use codeblocks myself, so I haven't been keeping the project files up to date; they probably need to have some of the new source files added to them (eg. new .c and .h files in src/renderer/). In term's of screenshots, here's one now. Keep in mind, this is a "manufactured" screenshot, in that I've just arbitrarily created some lights in order to show off the shadows. One is blue-ish and the other more orange so you can tell which shadows are being cast by which lightsource; lights in the actual game can be whatever color we want, though white (or close) is more realistic for most types of lightsource.
No lightmaps were used in this scene; in fact, I expect that the lightmaps will be removed alltogether by the time I'm done. In the mean time, keep in mind that this is still very much a work-in-progress; the renderer-work branch isn't really intended to be a viable alternative to trunk for gameplay or testing at the moment.
-
YEAH !!!
You are our OpenGL wizard 4 sure !!!
A triple 8) 8) 8)
& thanx for the hint with the codeblocks project file, I had no look @ that, it was just some quick testing...
Is it a big amount of work for you to try to fix the codeblocks project file ?!
My Linux OS is unavailable for me @ the moment, because everything is set up for high-res compiz-beryl-fusion gfx there, but my TFT has given up @ the moment & I do not want to rebuild my whole configuration...
Also fglrx in version 10.6 has brought lots of speed improvements, but still it is ATI :'( & you know how nice their drivers are, under Linux even more fX!!%!§...
But I have a GTX260 winxp machine waiting for a test of some real-time shadows & would be very happy to turn it on, whenever you are ready to give us a chance to test this...