I chuckled earlier this week when I saw an IBM TV ad talking about its petaFLOP computer. The joke is that Stanford's Folding@Home distributed computing application for protein-folding research recently hit five petaFLOPS -- 5 quadrillion floating-point operations per second. For the past two decades, scientists have been manipulating proteins to better understand how the nanomachines, I mean cells, that make life work and how diseases manipulate cell processes.
Folding@Home (and sister Rosetta@Home) reached 5 petaFLOPS by harnessing the multi-CPU power of today's graphics cards: nVidia GPUs represent over 2.18 petaFLOPS of computing power on only 18,333 computers. The future will bring real scientific advances through the raw, brute computation capability of distributed clients. Now, if we could only get a 24 hour weather forecast for New England with better than 75% accuracy (1965 accuracy was 74%), but that's a NASA problem.
If you leave your computer on most of the time, I urge you to experiment with FAH or Rosetta. It's a lot of fun tinkering and you are contributing to science (and heating the room).
Full disclosure: yours truly is #2 contributor at the #5 team on Folding@Home. And that's with an immature folding driver on an ATI graphics card.