Should I use POT size texture for render targets?

Hi, I’m working rendering pipeline where there are many render targets.
The whole rendering pipeline took a lot of render passes.
Each of RenderTarget needs to be input for another render pass.

I’m currently curious about POT texture size optimization.
Is it going to be better if I force the whole render target texture size to be POT?
Is it still relevant in today’s GPU?

Why do you think using NPOT would produce a performance hit? Most POT/NPOT topics in the WebGL space are related to WebGL 1 where NPOT texture has limited feature support (you can only generate mipmaps with POT textures). With WebGL 2, that is not relevant anymore.

1 Like

As @Mugen87’s comment suggests, WebGL-capable devices that cannot support NPOT textures at all are pretty much extinct, I would not worry about that. The question, “Do you need to support WebGL 1?” is more important. If you cannot require WebGL 2, and you need to do mipmapping, then using POT would still be important.

1 Like