CVAs problem

This is how I use CVAs:

for (i=0; i<numtex; i++)
if (SGE_driver->CVASupported())

But I do not see any speed improvement. Am I making it in the right way?

I have a GeForce2 Pro 64Mb

Thanx in advantage.

Nope, you are using it incorrectly. You need to lock the arrays, draw each pass, then unlock the arrays. But since you are not using a multipass algorithm, locking the arrays may actually slow you down a bit.

So, CVAs are only useful if I have to redraw anything multiple times per frame, isn’t it? such implenting lightmaps per software: first render the base texture geometry and then render the lightmaps, right?

Sure, that could be an example of how to use CVA’s if you couldn’t use multitexturing to combine the texture and lightmap in a single pass.

And are Vertex Array Range only useful if I have to re-render things multiple times per frame? like CVAs? or will it speed-up rendering any kind of things, having or not to re-render???

I’m not sure if you’ll understand me.
Excuse my English