View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001821 | FSSCP | OpenGL | public | 2008-11-16 19:32 | 2010-12-09 20:18 |
| Reporter | Tolwyn | Assigned To | Zacam | ||
| Priority | normal | Severity | crash | Reproducibility | always |
| Status | closed | Resolution | no change required | ||
| Summary | 0001821: GLSL Shaders cause an Access Violation in module atioglxx.dll | ||||
| Description | Game freezes/crashes to desktop (depends on Windows distribution) when a player select a complex model in tech room/the lab and specular, glowmaps or normal maps are enabled. If those maps are disabled or if GLSL (Shader) support is turned off everything is fine (despite lower graphics quality of course). Tested on several systems (1900XT, 1950XT) with different Catalyst versions. Older Catalyst (june) do not initialize GLSL shaders: shine mapping looks different and normal mapping does not work. New driver set (september and november) cause an Access Violation in module atioglxx.dll. Build used: svn build (21.10) and cutscene build by karajorma (30.10) game_flags = -ambient_factor 90 -wcsaga -fov 0.65 -spec_exp 7.0 -spec_tube 5.0 -spec_point 8.6 -spec_static 3.0 -spec -glow -env -mipmap -normal -ship_choice_3d -wcsaga -window ; | ||||
| Tags | No tags attached. | ||||
|
2008-11-16 19:32
|
|
|
|
The first test was with the WCS data, I just tried to reproduce the bug with the latest FSU data set. Same issue: most models crash the game in the tech room. --------------- fs2_open_3_6_10r-20081118_r4951 caused an Access Violation in module atioglxx.dll at 001b:696d5723. Exception handler called in FreeSpace 2 Main Thread. Error occurred at 11/18/2008 20:55:44. D:\Games\Freespace 2\fs2_open_3_6_10r-20081118_r4951.exe, run by Tolwyn. 2 processor(s), type 586. 2047 MBytes physical memory. Write to location 00000000 caused an access violation. |
|
|
Thought I'd pitch in here: I'm getting this too. ATI Radeon x1950, Catalyst drivers version 8.10 . Under Windows XP and Windows 2003, the game seems to hang indefinitely upon loading a model with normal maps in the techroom. No actual crash in Windowed mode. |
|
|
Got the fs2_open.log? |
|
2008-11-20 17:50
|
|
|
|
Log added. EXE used: fs2_open_3_6_10d-20081118_r4951 Commandline: -mod mod_cockpit -spec -glow -env -mipmap -normal -3dshockwave -ship_choice_3d -3dwarp -pos -window FSU data: 2-3 weeks old, checked out from svn |
|
|
Shaders (amongst other things) have been updated since this issue was created. I recommend updating from SVN and trying again. I presume you have the general read-only access information for it. If anybody else is interested in testing the MediaVP stuff, PM me on HLP. |
|
|
I've tried different combinations of default features using FSU latest dataset (advanced effects disabled) with some interesting results. glow maps disabled: usual behavior shine maps disabled: crash on entering tech room normal maps disabled: usual behavior env maps disabled: usual behavior glow & spec disabled: usual behavior spec & normal disabled: usual behavior glow & normal disabled: crash on entering tech room glow, normal and env maps disabled: crash occurs at a later point. I can load a few models that crashed the game before. specular, glow and normal disabled: everything works |
|
|
While I still don't have a x1950, I did get to borrow a friends x1800 XL. And I was unable to reproduce this issue with the limited amount of time I had it. I used Catalyst 8.10 and 8.11. C:\Games\FreeSpace2\fs2_open_3_6_10r.exe -mod mediavps -spec -glow -env -mipmap -nomotiondebris -missile_lighting -normal -3dshockwave -cache_bitmaps -dualscanlines -targetinfo -orbradar -rearm_timer -ship_choice_3d -weapon_choice_3d -3dwarp -snd_preload -loadallweps I did notice though that in my error log that the vertices and Indices matched in count, and they do not in tolwyns. Backslash, could you include a full fs2_open.log please (create if you have not already a file called debug_filter.cfg)? Either here or PM it to me. |
|
2008-12-11 23:45
|
|
|
|
Sorry about that... was having some additional problems and decided to wait until the 3.6.10 pack finished downloading. Well, that seems to have fixed the additional problems but the glsl problem is still here. Here's my fs2_open.log Interesting: Unlike most crashes, this 'hang' results in almost no CPU usage. |
|
|
Are you sure it's not just regenerating the cache files? I guess if it's not using the CPU that's unlikely but some models can take minutes even on new hardware to cache. |
|
|
Hmm, could be. Does it need to? This is with the just released 3.6.10 mediavps, including mv_cache (and I cleared my old cache folder)... are the cache files different whether or not 'disable GLSL' is on? Just in case: like Tolwyn, if I disable GLSL in the launcher, no hang occurs. It'd be one thing if the x1950 wasn't supposed to be able to support normal maps and stuff like that... |
|
|
GLSL shouldn't have anything to do with cache generation, but then again disabling it on a Mac should revert to the same behavior as 3.6.9, but it doesn't, so who knows. |
|
|
Okay. A hell of a huge headache later and some headbanging with folks _seems_ to narrow this down. Ish. It seems _some_ x1xxx cards have success (screenshot proof pending) by using the ATi OpenGL driver from the 7.1 Catalyst drivers package. Prior to this though, the idea was reached (spawned by the idea of caching and purging) that ATi's SM3.0 support on these Pre-HD2xxx is flat out flawed/broken due to not supporting _Vertex_Texture_Fetch, but must instead rely on _Render_To_Vertex_Buffer. But I don't know if this idea has any legitimacy or not and I won't know until I have both card's, code and some vacation time. This _may_ explain the issues odd behaviour in regards to what does and doesn't set it off and how it exhibits it's failure. VTF takes place on nVidia cards by the support and execution of GL_NV_vertex_program3 in case anyone is interested. But I am not sure (not yet having one of these cards or having seen screenshot proof) why an older ATi driver would work on something that ATi decided not to include because it was supposedly "too slow". (Actually, I have plenty of good ideas, but here is not the place for them.) |
|
|
Well, this has been here a while. With the advent of Antipodes changing some behaviour on shader handling, the following shaders were created (but they do operate, albeit with some side effects on .12 and Nightlies) http://www.hard-light.net/forums/index.php?topic=71890.msg1421074#msg1421074 Code in Trunk was lightly modified to remove the need for SM3.0 to operate GLSL (by The_E commit 6353) which should allow the shaders linked above to operate on ATI hardware, including with -normal maps. However, the full range of cards it will actually do that for is still unknown. |
|
|
As this is an issue with the Catalyst drivers, there's nothing much we can do here. IMO, this is a support issue more than anything. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2008-11-16 19:32 | Tolwyn | New Issue | |
| 2008-11-16 19:32 | Tolwyn | File Added: errorlog.txt | |
| 2008-11-18 19:57 | Tolwyn | Note Added: 0010225 | |
| 2008-11-20 04:18 | Backslash | Note Added: 0010237 | |
| 2008-11-20 16:22 | chief1983 | Note Added: 0010239 | |
| 2008-11-20 17:50 | Tolwyn | File Added: fs2_open.log | |
| 2008-11-20 17:51 | Tolwyn | Note Added: 0010240 | |
| 2008-11-30 08:20 | Zacam | Note Added: 0010303 | |
| 2008-11-30 17:31 | Tolwyn | Note Added: 0010308 | |
| 2008-12-07 03:00 | Zacam | Note Added: 0010381 | |
| 2008-12-07 03:00 | Zacam | Note Edited: 0010381 | |
| 2008-12-11 23:45 | Backslash | File Added: backslash_fs2_open.log | |
| 2008-12-11 23:49 | Backslash | Note Added: 0010398 | |
| 2008-12-11 23:52 | chief1983 | Note Added: 0010399 | |
| 2008-12-12 03:59 | Backslash | Note Added: 0010401 | |
| 2008-12-12 06:15 | chief1983 | Note Added: 0010402 | |
| 2008-12-17 05:56 | Zacam | Note Added: 0010424 | |
| 2009-11-10 16:58 | Zacam | Status | new => assigned |
| 2009-11-10 16:58 | Zacam | Assigned To | => Zacam |
| 2010-10-06 17:06 | Zacam | Note Added: 0012391 | |
| 2010-10-06 17:08 | Zacam | Status | assigned => feedback |
| 2010-12-09 20:18 | The_E | Note Added: 0012521 | |
| 2010-12-09 20:18 | The_E | Status | feedback => closed |
| 2010-12-09 20:18 | The_E | Resolution | open => no change required |
