Why just one processor?
Do you really have to cram it all together?
Take a scope for example, If you think of overlays, you could have some parts of the screen be updated by the gui processor and some of that could be bit map pictures out of the fpga.
Say for example you built a scope, lets go big time and use a HDMI display and pushing it to HD limits just to make it harder.
Say every thing on the front panel but the bnc connectors was part of the GUI processor section. So the GUI is handling the touch screen, knobs, buttons & switches.
The BNC with it high speed requirements really needs a FPGA and once the data is in the FPGA's memory really could use a lot of parallel processing. It would take a very slow data rate for the GUI section to pass the needed control information to the FPGA section and it could pass along some more data saying put trace 1 in this window on the display.
The GUI section could output Full HD and send it to a simple video overlay processor, a part of the FPGA section which would do the simple thing of using GUI data for all but the defined windows where it would use the the data created by the FPGA section, the wave form.
Think of this you could stack a second board behind the scope's FPGA and have a second instrument or a second scope fpga system able to add it's traces to data windows to where they were needed if they were needed at the time.
Want a 8 channel scope, just stack 4 two channel boards. Granted stacking scope boards would need some communications between scope boards for triggers and sample timing. Chanels 1 & 2 would have a hard time effecting any thing but 1 & 2.
The GUI board could be a common part used in many different equipment.
Want to use a normal PC & the GUI board? You would have two HDMI's that you would need to use a fancier overlay processor that had sync capability.
C