Basically the technology seems to work like SoftTH (Software Triplehead), only in an almost reverse fashion.
I'm almost surprised it's taken this long to get done, since the basic premise is so obvious that it's already been done before of a fashion, just not applied to this area.
SoftTH renders a scene on one card, and then splits it up and sends part of the frame to a second card for displaying on extra monitors above and beyond what the rendering card can output to. In this case you are just sending all the data to the "other" card and have it do the displaying while the power card just renders.
Sure, they have added in a copy buffer, but the basic premise of sending the frame buffer to the system RAM to be displayed by a different card to that rendering it has been done previously.
The obvious advantage of doing it with IGP/in mobile systems is the fact that the IGP is using the system RAM as a framebuffer, while in SoftTH, the frame buffer gets sent from card A to the system RAM to card B and then gets displayed, resulting in issues due to bandwidth use etc.
(Plus they obviously added dealing with powering up the card for specific applications so it doesn't have to work all the time).
The fact that it's an already proven rendering solution though means that hopefully AMD will also adopt a similar elegant solution which will mean that you can buy any discrete laptop and be able to use it with any IGP.
It's a bit of a shame NV has decided to only allow their solution to work with Intel and NV IGPs for now, and time for "platform" company AMD to up their mobile game.