Behind glVertexArrayRangeNV


I wrote a landscape renderer using dynamic quadtrees. For now I use normal Vertex Arrays for each Node wich consists of 9 Vertices. Now I’m trying to improve the performance using VAR extension.

I thounght it would be a good idea to take more than just 9 vertices per call to glDrawElements but I don’t really know what VAR ist doing.

If I use a Vertex Array of 81 Vertices that lies in AGP memory but glDrawElements touches only 9 out of these 81 would all 81 be pulled by the graphics card or just 9? I could also imagine that the part of the array from the lowest to the highest index of the 9 vertices would be pulled (what would be very bad for my implementation).

Any idea what would happen ?

[This message has been edited by muhkuh (edited 04-24-2002).]

VAR works exactly like normal vertex array, except that it pulls vertices using DMA.

[This message has been edited by Gorg (edited 04-24-2002).]

You can probably write an alogrithm to produce a tri-strip (in real time, of course) of the landscape to make it go faster too…Just a suggestion.