Instead of using one fast processor to get a task done, Carnegie Mellon University researchers have set up a system of slower processors that can do the same work in about the same time, using less energy. The structure the researchers have found is called "Fast Array of Wimpy Nodes."

The CMU scheme is basically a parallel computing idea that relies on programmers to decide how different low-power processors with local Flash storage are responsible for what sections of information. Researchers say the structure could cut datacenter energy costs in half, though it is a bit time intensive for the programmers involved.

The CMU researchers recently published an ACM paper on their work:

http://www.cs.cmu.edu/~fawnproj/papers/fawn-sosp2009.pdf