View Full Version : hardware accleration
04-22-2000, 06:52 PM
hey i was wondering how you would enable hardware accleration with say a TNT2. any help would be greatly appreciated
If you use Microsoft´s OpenGl-Implementation
you don´t have to enable Hardware-Acceleration,´cause this implementation uses a 3d-Accelerator automatically if there is one.
If you use the SGI-Implementation you can´t get Hardware-Acceleration.
P.s.:I assume you´re using Windows.
04-27-2000, 05:31 AM
You can indeed get hardware acceleration with the SGI implementation ...
04-27-2000, 12:20 PM
How that's possible?
You ever did it?
It indeed never happened to me.
Actually, I use SGI OpenGL if I want to test code with software rendering.
just courious about that... how do you switch, on the same physical system i mean, the opengl drivers to render in hardware or in software?
personally, i tweak the registry, wich works on win32 and nt...
04-27-2000, 11:03 PM
You can do it in the application via the SetPixelformat() call.
Check out the documentation on the PIXELFORMATDESCRIPTOR. Look for dwFlags, it must contain PFD_GENERIC_FORMAT. (If it has PFD_GENERIC_ACCELERATED set, you'll might run on a Mini Client Driver (MCD)).
Check also the GL_VENDOR string once you got the OpenGL context running. If it says "Microsoft Corporation" and not your hardware vendor's name, you have the GDI Generic Software Implementation running.
Selecting a software pixelformat can also happen via ChoosePixelformat() with options the hardware does not support (no stencil buffer is one of the favorite reasons). The Microsoft formats offer different color, z, and stencil buffer depth as fallback.
04-28-2000, 01:05 AM
> You can do it in the application via the SetPixelformat() call.
But you cannot do it without source code modification...
> personally, i tweak the registry, wich works on win32 and nt...
Registry? hmm, it's depend..
In office I have 2 PC:
with FireGL1000 (Permedia2) and with Elsa Gloria L/MX (GLINT MX).
Driver for L/MX has registry key ICDInterface.Disable, don't know about FireGL1000.
I use simplest way: I rename SGI's opengl.dll to opengl32.dll and copy to directory with my exe.
Usually it works better then MS software-only implementation, and texture mapping indeed faster.
Also I found that SGI dll can use hw acceleration - BUT only if application is compiled to use opengl.dll, not opengl32.dll
It look like SGI's dll searches for opengl32.dll in the current directory, and finds nothing but itself.. http://www.opengl.org/discussion_boards/ubb/rolleyes.gif
It makes this ability hardly useful.
Powered by vBulletin® Version 4.2.3 Copyright © 2017 vBulletin Solutions, Inc. All rights reserved.