OptiX rendering, to my knowledge, uses the RT cores in the Turing architecture cards like my 2060. How does tile size affect the render speed?

As far as I know the general rule is to increase tile size for GPU rendering and decrease for CPU rendering. I did not notice a significant difference in render times between 64x64 and 3840x2160. I'm sure there are a million things wrong with this testing, but I have no idea what they are. I don't think I know enough to test and pull off a rigorous test, so I was wondering if there is a benchmark my simple google search can't find or (if such a benchmark is unavailable) there are guidelines to test this myself on my own hardware.

Thanks for any help you can provide!

Edit 1

I did some quick testing and I'm beginning to think that larger tiles are faster, but I'm still not sure. I have no idea how to test things like this well and rigorously, so I think I'll just go with the rule of thumb bigger tile size is better... I'd still appreciate any advice on running tests myself or seeing someone else's tests.

  • $\begingroup$ Hello :). Making tiles larger helps only to some tile-size, after which the speed starts to decrease again. You can make us of the Auto Tile Size addon, that comes with Blender, but I recommend you spend one evening testing different tile sizes. I get best results with 480x360 for GPU rendering. Anything larger, and the speed starts to drop. $\endgroup$ – Jachym Michal Dec 2 '19 at 16:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.