Sad that as usual everything descends into a dustup to the detriment of useful conversation. I waded through seven pages and found perhaps a half page of intelligent discussion.

I for one love the idea of a CPU-based PBR, although long-term I hope to have better GPU options. My computer itself is incomparably more powerful than my video card.

The interesting thing I've noticed is that people are howling for GPU rendering for Cycles without actually looking at what Cycles GPU rendering can do.

In the current release, there are significant differences between the capability of GPU rendering on NVIDIA cards and AMD cards. There's also differences between what can be done on the CPU and on the GPU.

Want to do subsurface scattering? Available on CPU rendering. Experimental on CUDA. You're out of luck in OpenCL. Want to do smoke or fire or volume? Well, the volume you can do on the CPU and CUDA cards. Out of luck for AMD again. And for fire and smoke, it's not happening on your video cards at all.

If I were in the process of adopting the Cycles engine, of course I'd focus on the CPU implementation. Not being able to do scattering? Yeah, that doesn't work real well for a software program designed primarily for human figures combined with a render engine that's supposed to be for realism.

One almost gets the impression that people have heard "GPU rendering" and think it's a magic term and thus there's no need to look further and see what exactly that means.