Top Ten Uses For A Mac-Based Supercomputer
Seriously, when you're doing research in things like "quantum chemistry, computational biochemistry, and cell cycle modeling" (--MacWorld), you need this kind of power.
When I was in engineering at UofT, some of my T.A.'s mentioned having to wait months to get time on supercomputers for their thesis work so they could simulate things like turbulence for their aerospace stuff.
In 1926 Erwin Schrodinger discovered how to calculate the energy levels of the Hydrogen atom. The discovery was so powerful that it could, in principle, be applied to any atomic or molecular system.
Problem is, for any atomic system with more than one electron (every other element) or with molecules, which are made of 2 or more atoms, you cannot solve the equations, but only make approximations.
The nature of these approximations lend themselves to computers, because the more iterations of the approximation you can make the better the approximation becomes, but the number needed to get even acceptable approximations for mid-small sized molecules quickly gets way out of hand.
So now to a contete example:
Say you want to find a really nice approximation to the molecular geometry of a section of DNA (what I mean by "really nice" is an accurate flourescence spectrum). You are then talking about hundreds if not thousands of atoms all involved with different chemical bonds which are goverened by the laws of Quantum Mechanics first shown my Mr. Schroedinger. To do that we are talking serious CPU time.
A random google example calculated on the world's 2nd fastest cluster (the G5 is #3).
I hope that gives you a glimpse of how much we need computing power to do interesting and important science.