PDA

View Full Version : .NET applications conflict with OpenGL games in current Forceware versions


Grestorn
10-02-05, 12:35 PM
After someone noted a noteable drop in framerates in Doom 3 if nHancer is running, I made some tests.

I could reproduce that behaviour by using Doom3, ("timedemo demo1 usecache"). With SLI I get 85 fps, but if nHancer is also running in the background, the rate is almost cut in half to 44 fps.

Even without SLI I get 55 fps and again 44 fps with nHancer running.

Since I couldn't imagine what nHancer could possible do to cause that effect, I made some additional tests. I found out, that it's not really neccessary to run nHancer at all, most .NET applications show the same effect.

For example, you can try the DXTweaker (available here (http://www.nonatainment.de/portal/DesktopDefault.aspx?tabindex=7&tabid=19)), a tool that's supposed to add some features to Direct3D games.

If I just start the DXTweaker without using it for any game, Doom 3 is immediately limited again to 44 fps on my system. As soon as I quit DXTweaker, the frame rate is back up again.

I can even reproduce that effect with the plain Visual Studio .NET 2003. Just open it, create a new C# Windows Form project, and then as long as the Studio is running, Doom 3 is slowed down.

I'm pretty sure .NET does something that limits the OpenGL performance in the current Forceware versions. Some conflict or a resource that .NET is allocating, I don't know.

Anyway, that's something that has to be investigated. .NET applications are bound to be more and more common in the future, and it would be a great nuissance if you had to quit all .NET apps before you use any OpenGL game.

BTW, Direct3D games don't seem to be affected by this at all. And not all .NET applications show this effect either. I think it's somehow connected to the way Windows Forms are created through the Visual Studio.

Again my request if someone with connections to the driver development team could please forward this for further investigation?

rewt
10-02-05, 01:05 PM
I'm not having this problem with .NET 2.0 beta 2. In fact my second bench in Doom3 was higher when I had my .NET 2.0 application loaded. I also used cache mode on both benches.

Grestorn
10-02-05, 01:12 PM
I'm not having this problem with .NET 2.0 beta 2. In fact my second bench in Doom3 was higher when I had my .NET 2.0 application loaded. I also used cache mode on both benches.
Well, it would be great if .NET 2.0 wouldn't show this problem, but as long as this is still beta, I can't really use it for nHancer.

And most .NET software is also still using 1.1 unfortunately... :(

rewt
10-02-05, 01:18 PM
Most 1.1 applications should still run as long as the user has installed .NET 1.1 prior to installing .NET 2.0.

Grestorn
10-02-05, 01:24 PM
Most 1.1 applications should still run as long as the user has installed .NET 1.1 prior to installing .NET 2.0.
Certainly, any .NET framework can coexist with any other framework version.

Still, I can't use 2.0 yet, especially since there's a component I use which is not yet fully 2.0 compatible.

Could someone please confirm that problem with .NET 1.1 (using DXTweaker for example)?

rewt
10-02-05, 01:28 PM
What component do you speak of?

Grestorn
10-02-05, 01:37 PM
What component do you speak of?
This one (http://www.blackbytes-components.com/product.aspx?id=2). I need it since the internal .NET Registry classes doesn't support adding change notifications on specific registry keys. nHancer needs that to automatically recognize any changes made through the control panel or any other tweaking tool.

rewt
10-02-05, 01:56 PM
Gotcha

When you say current ForceWare versions does the issue occur on the ForceWare 81.xx as well? Another question I would ask is could it be specific to Athlon 64 machines?

rewt
10-02-05, 02:21 PM
I uninstalled .NET 2.0 beta 2, which left me with .NET 1.1 sp1.

Ran Doom3 several times with .NET applications unloaded& loaded (including DXTweaker and nHancer 1.2.2) and could not recreate the issue you are experiencing. Take that for what it's worth :-/

Athlon XP-M 2500+, Geforce 5900, Forceware 81.82

Grestorn
10-02-05, 03:01 PM
Well that's strange indeed, since somebody else was able to recreate the problem.

I tried it with 78.03 and with 81.82. My system components are listed in my signature. I'm using normal Win XP Pro 32 bit

Well, at least it seems that not everybody is affected, I guess that's good news... even though it might make it more difficult to track it down.

rewt
10-02-05, 03:04 PM
I'm not so sure it'll make it harder. ;)

Are all the people having issues using 64-bit processors on 32-bit Windows?

Grestorn
10-03-05, 03:48 AM
Please, I need more data to track this down.

Could some people please try to benchmark Doom3 with either nHancer or DXTweaker running and without any of these tools and post the results?

I would very much appreciate that!

Zandig
10-03-05, 10:24 AM
Just tried it using "Timedemo demo1 usecache"

with nHancher running my frame rate was indeed cut in half!
closing nHancer and running the test again frames back to normal.

However if I just ran a "Timedemo demo1" frame rates remained normal with nHnacher running or not.

Hope that helps.