I don't think pixel shaders of any revision are going to help framebuffer emulation. Though to be honest I don't know what tricks orkin and Rice are using to speed up the effects. The limitation, as I understand it, is in PC architecture, the AGP bus specifically. It just wasn't designed for the multiple reads and writes required by framebuffer effects. I'm curious to see if PCI Express will help.
In the meantime, there's always the software approach. A software plugin will perform framebuffer tricks just fine, but, of course, has its share of trade-offs in other areas.