I’ve written a small app which renders various models using either individual triangle calls (glBegin(GL_TRIANGLES)) or compiled vertex arrays. Now on my TNT I get the expected behavior, the compiled vertex arrays are a good deal faster. But on a another machine with a faster processor and a TNT2, using the compiled vertex arrays actually makes it run a go deal slower. Also of note, is that on that machine Q3A runs much faster than Quake 2. I seriously doubt this is normal behavior for the TNT2 machine, but I was just wondering if anybody has seen this problem before and what could be causing the trouble.
P.S. I’ve tried old and new drivers on the TNT2 machine with no difference.
I’ve seen a few things like this on one or two TNT2’s. I’ve encountered a few hiccups when trying to get high end 3d packages such as Maya to run on them. TNT2’s actually cause Maya to bomb out and crash, this being a package that’ll happily run in software mode if 3D capabilities of your card aren’t up to scratch.
I don’t know exactly what the problem is or how to solve it… but from a process of elimination I came to this conclusion…
TNT2’s don’t support all the function calls in openGL 1.1 (not sure if the new drivers support 1.2, but they didn’t last time I used the card), or there is a fatal error generated when some of those functions are called. I think vertex array’s are very much a problem with the cards. That’s all I know, sorry I can’t be more specific.
[This message has been edited by Rob The Bloke (edited 06-21-2000).]