I can see how higher resolutions will require the GPU to have more memory or more FLOPS or both. I guess my question is, how, if at all, are the improvements required to support 4K different from improvements that target something like deep learning performance?