Just to gently interject, a De Laval nozzle only works as stated if the input air velocity it sufficient to achieve sonic speeds when "pinched" through the choke point, thus resulting in a supersonic exhaust expansion on the far side. An "impeller in the low pressure choke point" doesn't make any sense; a De Laval nozzle doesn't "suck", it "blows". By the very definition your source of pressurized gas has to be upstream of the entire hourglass for your device to be considered a De Laval nozzle.Putting an impeller into the low pressure choke point of a De Laval nozzle is the mostthing I've seen in a case design.
The G5 cooling system may possibly leverage the Venturi effect to "vacuum" heat away from the CPUs and other hot parts (something shaped like a De Laval nozzle acts like a Venturi tube at speeds lower than the critical intake velocity), but I seriously doubt that the compressibility of air at sonic velocities plays into the design much. I remember G5s being "loud" when the fans were turned up to 11, but even the most powerful vacuum cleaners don't pull more than a couple PSI of vacuum. (It's basically impossible for a single-stage fan to do so under normal atmospheric conditions.) For the G5 case to act like a De Laval nozzle you'd probably have to aim the ouput of a good-size truck supercharger* through the intake side and I have no doubt the 700+ MPH exhaust blowing out the back would have some... negative side effects. (Kinda think the first thing to go would be the blades of that "impeller sitting in the choke point". Followed simultaneously by the plaster on the wall behind your computer desk and the thin membranes inside your auricular canals which we call "eardrums".)
(Edit: Actually, without getting into math that's way over my head I can't make more than the loosest estimate about how much compression you'd need upstream, but thinking about it again and just eyeballing the size of the fans and degree of "pinch" there is in the G5 case I revise my estimate upwards to thinking you'd literally need to aim the blast of a turbojet engine like a J-47 through it to get supersonic exhaust on the far side. That or, of course, a combustion chamber using the rocket fuel/oxidizer mix of your choice. There just isn't that much "pinch" though, so it's really not going to make for a very good supersonic nozzle. In the case of the J-47 suggestion I'm cheating and including the funnel you'd have running from the back of the engine to the case in the intake throat.)