Graphic symptoms
by PeterB · in Torque Game Builder · 08/28/2007 (6:27 pm) · 9 replies
I have two symptoms that could be from either bad practice or hardware problems. I am testing my prototype on various machines and I found one that has problems.
With my game, after pressing play, I see black for 5 seconds while everything loads. Cool.
But on my 2.5 year old laptop specifically, I see these kinds of artifacts:
www.pabit.org/temp/
Or if I previously had opened and ran a different TGB project, and then return to my problem project, I'll see a jumbled image of the previous project.
The game runs fine. This sounds like something to do with video buffer?
And even after I completely close Torque, windows minimize and maximize slowly. The video card in general seems to perform badly until I restart. This does not happen with other Torque projects (which have smaller images I admit).
Playing commercial 3D games give inconclusive results- I have seen a couple of tiny artifacts that were not there a month ago. I think.
In the Torque project, I do have over 10 images that are 1024x1024, which I know is pushing my luck. On a fast machine it causes no problems. This laptop has 128MB of video ram.
A test- after removing all large images, there is no performance hit, but I still get the crazy artifacts while the game is loading.
Until I find another game that gives similar results, I won't know if it's my project or my (dying) video card. Any input would be appreciated.
With my game, after pressing play, I see black for 5 seconds while everything loads. Cool.
But on my 2.5 year old laptop specifically, I see these kinds of artifacts:
www.pabit.org/temp/
Or if I previously had opened and ran a different TGB project, and then return to my problem project, I'll see a jumbled image of the previous project.
The game runs fine. This sounds like something to do with video buffer?
And even after I completely close Torque, windows minimize and maximize slowly. The video card in general seems to perform badly until I restart. This does not happen with other Torque projects (which have smaller images I admit).
Playing commercial 3D games give inconclusive results- I have seen a couple of tiny artifacts that were not there a month ago. I think.
In the Torque project, I do have over 10 images that are 1024x1024, which I know is pushing my luck. On a fast machine it causes no problems. This laptop has 128MB of video ram.
A test- after removing all large images, there is no performance hit, but I still get the crazy artifacts while the game is loading.
Until I find another game that gives similar results, I won't know if it's my project or my (dying) video card. Any input would be appreciated.
About the author
#2
I must admit that I got rid of the performance problem by updating the bios. But I still get the artifact. So it's not the end of the world, I must admit.
08/30/2007 (4:39 pm)
Yes a 128MB has trouble.I must admit that I got rid of the performance problem by updating the bios. But I still get the artifact. So it's not the end of the world, I must admit.
#3
Getting the most current version of the drivers helped.
But it might actually be an issue with VRAM.
What Graphiccard do you have in your notebook?
How much VRAM?
Do your images use padding? because that will raise your VRAM need as it splits up the 1024x1024 ... (which themself only would be 40-50MB of VRAM!)
As well the amount of RAM on the machine might have an impact there just to mention.
But there is one thing you did not mention: did you check the console log if the OpenGL mode was started or not?
On DX mode, it will perform considerably worse I fear as TGE is focused around OpenGL and DX is just a "wrap over" fallback. And that fallback even has some known limitations like no 3D shapes.
Without having any further hints on the hardware and driver versions, its hard to say what might be the exact cause of the problem beside what you already assume: some kind of backbuffer issue.
09/01/2007 (6:03 am)
I got that kind of error when I had outdated drivers on my Intel GMA900 onboard in my tablet. (especially when using vista)Getting the most current version of the drivers helped.
But it might actually be an issue with VRAM.
What Graphiccard do you have in your notebook?
How much VRAM?
Do your images use padding? because that will raise your VRAM need as it splits up the 1024x1024 ... (which themself only would be 40-50MB of VRAM!)
As well the amount of RAM on the machine might have an impact there just to mention.
But there is one thing you did not mention: did you check the console log if the OpenGL mode was started or not?
On DX mode, it will perform considerably worse I fear as TGE is focused around OpenGL and DX is just a "wrap over" fallback. And that fallback even has some known limitations like no 3D shapes.
Without having any further hints on the hardware and driver versions, its hard to say what might be the exact cause of the problem beside what you already assume: some kind of backbuffer issue.
#4
I wiped my HD and installed all the newest drivers.
1.25GB of regular ram
Opengl seems good in the console.
What do you mean by padding on the images. Please explain what you mean here: "Do your images use padding? because that will raise your VRAM need as it splits up the 1024x1024 ... (which themself only would be 40-50MB of VRAM!)"
Since my project is simple, and I definitely didn't touch the source, would a 'backbuffer' problem definitely be hardware, not poor development on my part?
09/01/2007 (2:39 pm)
Radion 9600 turbo. 128MBI wiped my HD and installed all the newest drivers.
1.25GB of regular ram
Opengl seems good in the console.
What do you mean by padding on the images. Please explain what you mean here: "Do your images use padding? because that will raise your VRAM need as it splits up the 1024x1024 ... (which themself only would be 40-50MB of VRAM!)"
Since my project is simple, and I definitely didn't touch the source, would a 'backbuffer' problem definitely be hardware, not poor development on my part?
#5
While this is a nice effect, you raise the VRAM usage of your graphics as TGB will split the 1024x1024 in subimages and repackage them depending on your settings (size or speed favor) which causes more or less extra usage above the original 6MB RAM / VRAM used for the texture ... (1024x1024 + mipmap is sadly around 6MB of VRAM)
I so far just use smaller image cells if I use cell based animation images. This means instead of 128x128, they are 126x126 but aligned in 128 x 128 cells with 1 pixel border around them. that removes the need for filter padding and allows me to control the texture rebinds instead of relying on a system I do not fully understand yet (or fully trust in all situations)
But that you have ATI might be another thing that could cause potential problems.
Did you try to set it to DX and see what happens?
If you are on Vista: Forget ATI + OpenGL for the moment. The drivers are massively broken ... Don't know if AMD is really working on the problem, even Intels Cards work better and thats no good sign I think.
09/02/2007 (2:24 am)
With padding I mean: if you load images in cell mode, you can enable filter padding to remove the clamping artifacts.While this is a nice effect, you raise the VRAM usage of your graphics as TGB will split the 1024x1024 in subimages and repackage them depending on your settings (size or speed favor) which causes more or less extra usage above the original 6MB RAM / VRAM used for the texture ... (1024x1024 + mipmap is sadly around 6MB of VRAM)
I so far just use smaller image cells if I use cell based animation images. This means instead of 128x128, they are 126x126 but aligned in 128 x 128 cells with 1 pixel border around them. that removes the need for filter padding and allows me to control the texture rebinds instead of relying on a system I do not fully understand yet (or fully trust in all situations)
But that you have ATI might be another thing that could cause potential problems.
Did you try to set it to DX and see what happens?
If you are on Vista: Forget ATI + OpenGL for the moment. The drivers are massively broken ... Don't know if AMD is really working on the problem, even Intels Cards work better and thats no good sign I think.
#6
I get the same types of artifacts when I run my Game from TGB. I am running TGB 1.5.1.
It looks like all my graphics all jumbled up for about 1 second while loading.
I am running on a Gen 1 XPS Laptop with a P4 3.4 and the ATI Mobility Radeon 9800; 256MB of VRAM.
I have 1 GB of RAM.
Other than the Jumbled Graphics while loading which looks BAD, I don't seem to have any other
issues.
09/02/2007 (5:25 pm)
Just for a little more input on these graphics issues.I get the same types of artifacts when I run my Game from TGB. I am running TGB 1.5.1.
It looks like all my graphics all jumbled up for about 1 second while loading.
I am running on a Gen 1 XPS Laptop with a P4 3.4 and the ATI Mobility Radeon 9800; 256MB of VRAM.
I have 1 GB of RAM.
Other than the Jumbled Graphics while loading which looks BAD, I don't seem to have any other
issues.
#7
I am using XP. I'm glad to know more about padding.
Dustin,
You have an impressive machine, so that teaches me something. Now that my computer is properly updated, I agree with your last sentence.
09/02/2007 (6:10 pm)
Marc, I am using XP. I'm glad to know more about padding.
Dustin,
You have an impressive machine, so that teaches me something. Now that my computer is properly updated, I agree with your last sentence.
#8
But only until the actual graphic data is loaded, similar to Dustin
From what I was able to tell so far it seems like it is caused by a too small initialized graphical backbuffer thats rendered as I was able to get desktop screen fragments and fragments from other 3D apps running while starting the app which looks like it reads backbuffer data from areas not intialized so far or better: not part of itself.
Haven't checked if the system suffers from this problems on 1.1.3 as well.
If not, a diff between 1.1.3 and 1.5.1 might give a hint where this problem was implemented.
09/03/2007 (2:20 am)
OK the loadup thing is a different issue. Get that effect on my GMA900 system as well (not able to get it on my desktop, but hardly a wonder. C2D @ 2x 3Ghz, 8800GTS OC, 2GB RAM, Raid 0 HD loads the app faster than it can open the window normally) and as it is quite slow it can see this fragmentation effect for quite some time.But only until the actual graphic data is loaded, similar to Dustin
From what I was able to tell so far it seems like it is caused by a too small initialized graphical backbuffer thats rendered as I was able to get desktop screen fragments and fragments from other 3D apps running while starting the app which looks like it reads backbuffer data from areas not intialized so far or better: not part of itself.
Haven't checked if the system suffers from this problems on 1.1.3 as well.
If not, a diff between 1.1.3 and 1.5.1 might give a hint where this problem was implemented.
#9
06/13/2008 (10:49 pm)
Sorry to dig this up from the grave, but one of my testers is having this problem with 1.72. Marc, if you're still around, did you have a code fix for this? 
Torque Owner Isaac Barbosa
IQ Games