<quote>The main reason to still keep an eye on POT dimensions is performance. NPOT textures are a bit slower, because the GPU needs to do more computations.</quote>
How much slower? Also, when exactly and why? - suppose I’m not using mipmapping nor using any sort of repeat wrapping, will my NPOTs still be slower?
Lets suppose I have a very large texture (1280x1024), and it makes sense to use an NPOT (because 2048x1024 would waste a lot of memory as padding) - will to be significantly slower to render this very large texture?
Simply saying that they are slower isn’t enough information to go on in making the decision.
If, by converting from my packed POTs to NPOTS, I’m trading in a bit of memory for the sake of performance, then that sounds like a very poor trade indeed.
Haha! I make this thread, then 20 minuates later I google for “opengl performance of npots” and this very thread is the first link! Google getting faster or what? :eek: