You'll see that the recent Kepler/Maxwell parts are single precision optimized which makes them more energy efficient (similar to past AMDs) but poorer at higher precision compute, in general
This is certainly true, and I have mentioned the weak DP performance of Nvidia GPUs before, but does it matter? What kind of OpenCL workflows are important to a con(pro)sumer? We are mostly looking at image and sound processing. None of these need double precision, even if you are doing HDR. What else? Cryptography? Here you need integer and bitwise operations.
From the top of my mind, only scientific stuff would need double precision. And you are not going to run that on a laptop. From general publics perspective, double precision is a neat feature that nobody actually needs.