We're talking about equivalent single-screen rendering requirements - how much work does the console need to do to render 60 Hz stereoscopic versus 60 Hz to TV. Geometry setup is resolution independent so every camera draw requires the geometry to be set up and 120 camera renders requires 120 geometry setups. Pixel drawing is resolution dependent and so 120 half-res draws is the same workload as 60 full-res draws. Thus to render a stereoscopic (VR) game at 60 fps, you have equivalent of 60 fps 1080p pixel drawing, but 120 fps geometry setup (regardless of res, but I called it 1080p120).
That's naive rendering. Real geometry workload can be reduced using Clever Tricks just as stereoscopic rendering.
