I guess that's pretty subjective as all supercomputers possess different computational power, usually a speed measure (Tflops/s or multiples of one trillion floating point operations per second).

To get what you're looking for you'd have to pick a particular supercomputer, IBM's Blue Gene for example, find its speed, and divide into it the speed of the average (whatever that may be) computer connected to the WCG. Then you have a co-efficient by which to divide the time figure detailed in the post by Uplinger. That would give you (more or less) the time it would have taken Blue Gene to do the same work.