lol I know what unified shaders meant.
I've been following tech sites since forever man. Aparently it's you who has not been around for too long and hence the jump to unified shaders means more than it really is, historically seaking. The introduction of shaders was an even more important change and maybe even more important was the introduction of hardware T&L. Unified shaders just means you can improve load balancing by running both vertex and pixel shaders on the same processor. It really isn't that big of a change in reality, coders have not changed their way of designing, they have not vastly increased the use of vertex processing or pixel processing in any direction. Current engines have about the same balance, compared to each other, there is not one with astronomic vertex processing and one wich only does pixel shading and in the end unified architecture just meant a more efficient use of the hardware, but didn't change anything.
What we have in Fermi is at least as innovative as that. We have an architecture optimiced for both graphics and GPGPU and will not make a big difference either for both, but it's the only architecture that permits doing both at the same time efficiently. It has also taken something that was suposed to be a fixed function running on dedicated hardware and has made it moldable, with expanded functionality. I'm talking about their aproach to tesselation.
All in all, either you overestimate the unified shaders or you underestimate all the changes in Fermi.