Over at the forums gamefish and BurningHand got a little worried about the supposedly bad performance of TextureAtlas. They did some benchmarking on G1/G2/Hero level hardware and were amazed to find that the benchmark could only achieve 6 frames per second when using TextureAtlas but a nice and niffty 60 frames per second using a Texture directly. Nate and me proclaimed strongly that there’s no difference in the renderpath when using a TextureAtlas or a Texture. They are indeed exactly the same. Sadly i could only find the time to figure out what’s wrong an hour ago and i wanted to share my findings.
To understand what’s going on we need two things: a test case and knowledge about how texturing works. Let’s start with the former.
...
No hay comentarios:
Publicar un comentario