Graphic corruption with ATI cards
by Peter Haffenreffer · in Torque Game Engine Advanced · 09/16/2008 (10:30 am) · 5 replies
Hello all,
I am experiencing a graphical corruption on ATI cards using TGEA. There are polys shooting out of the player.
This is happening on both Vista & XP machines.
Here's what I know:
-The problem is definitely happening when trying to render shadows.
-I turned on the debug version of D3D and was getting the following error when trying to render the shadows: Stream 0 does not have required number of vertices. This was getting thrown on mD3DDevice->DrawIndexedPrimitive in GFXD3D9Device::drawIndexedPrimitive()
I have looked over this thread: http://www.garagegames.com/mg/forums/result.thread.php?qt=67774
but that is dealing with OpenGL and TGE. It seems as if the Pref::ts::usetriangles script variable has been removed from TGEA, so I tried changing TSMesh::smUseTriangles in code to true, but still no luck.
Has anyone seen this before or have any suggestions?
I could really use some help from the GG staff here.
Thank you.
I am experiencing a graphical corruption on ATI cards using TGEA. There are polys shooting out of the player.
This is happening on both Vista & XP machines.
Here's what I know:
-The problem is definitely happening when trying to render shadows.
-I turned on the debug version of D3D and was getting the following error when trying to render the shadows: Stream 0 does not have required number of vertices. This was getting thrown on mD3DDevice->DrawIndexedPrimitive in GFXD3D9Device::drawIndexedPrimitive()
I have looked over this thread: http://www.garagegames.com/mg/forums/result.thread.php?qt=67774
but that is dealing with OpenGL and TGE. It seems as if the Pref::ts::usetriangles script variable has been removed from TGEA, so I tried changing TSMesh::smUseTriangles in code to true, but still no luck.
Has anyone seen this before or have any suggestions?
I could really use some help from the GG staff here.
Thank you.
#2
09/17/2008 (7:52 am)
I believe someone from GG needs to weigh in here. This is a show-stopper problem and has something to do with the Torque rendering code. We haven't touched it, and we have seen other people voice the same problem. 
#3
Problems on ATI GPUs are well known. I've discussed it 2 years ago with no result.
Sometimes a working shader on NV isn't working the same way on ATI.
Try with another Radeon, problems on 1900 are absent on 38X0/48X0.
09/17/2008 (11:14 am)
I believe you.Problems on ATI GPUs are well known. I've discussed it 2 years ago with no result.
Sometimes a working shader on NV isn't working the same way on ATI.
Try with another Radeon, problems on 1900 are absent on 38X0/48X0.
#4
http://forums.rampidgaming.com/showthread.php?t=5405&highlight=radeon
09/17/2008 (11:26 am)
Have the same issue with Radeon's in TGE. Here is a thread about it:http://forums.rampidgaming.com/showthread.php?t=5405&highlight=radeon
#5
We've actually tested it on several ATI cards, and it is showing up on all of them so far.
The errors mentioned in the first post that D3D is throwing are being thrown on NVIDIA cards as well. NVIDIA cards just don't show the corruption, but it really isn't any better since the corrupt data is still present and can cause crashes later or other oddities.
So, the problem really isn't that ATI shows the corruption, but the problem is that these errors are being thrown and nothing has been done about it.
I've been looking into it for about 12 hours now, and still no luck.
09/17/2008 (11:29 am)
Well, I'm glad someone else is seeing this.We've actually tested it on several ATI cards, and it is showing up on all of them so far.
The errors mentioned in the first post that D3D is throwing are being thrown on NVIDIA cards as well. NVIDIA cards just don't show the corruption, but it really isn't any better since the corrupt data is still present and can cause crashes later or other oddities.
So, the problem really isn't that ATI shows the corruption, but the problem is that these errors are being thrown and nothing has been done about it.
I've been looking into it for about 12 hours now, and still no luck.
Torque Owner Peter Haffenreffer
Default Studio Name
If I prevent the mesh->render() call inside of sgShadowProjector::sgRenderShape() for all Skinned Meshes, the bug doesn't happen and D3D does not throw the error mentioned in my above post.
This leads me to believe there is something wrong with the creation of the skinned meshes? I'm not entirely sure, but I'm hoping that this information can help others out there get to the bottom of this very serious issue.