this means ATI (catalyst) + Nvidia in same machine? (hybrid like windows)
More like OpenCL I think. You query available devices and create GL context per device. I am wondering that it took them so long to figure this out. The difficult part will be the draw buffer ownership/sharing I guess.
Now nVidia has done something right and this GL dispatch mechanism is a step in the right direction for using multiple GPU's that are made by different makers in the same system. This will definitely help with something like Optimus for example. A thumbs up rather than a middle finger this time for nVidia as I'm sure they still were smarting from that middle finger attack by Linus.