there’s a demo on nvidia with Vertex Array Range… it’s a yellow surface with lots of waves. They get the memory from wglNVAllocateMemory() and change the data every frame in the allocated memory. Isn’t that slow, can we do that? they allocate the memory with 0 frequency of writting and reading!!! What do they do to make that demo sooooooo fast?
I wrote this demo specifically to illustrate that you can modify AGP memory very fast if you write sequentially into the memory. Even if you’re not using interleaved arrays, this should still be fast.
The fact that I use a writefreq of 0 is kinda bad. The thing is, I know it doesn’t actually matter (today) what that number is, because it doesn’t change the type of memory you get. I should probably have chosen a higher write frequency.