Exaflop Computing

Only recently (it seems) processing power passed the petaflop threshold (that’s one quadrillion floating point operations per second), and true to Moore’s Law, it’s expected that we’ll be seeing exaflop (one thousand petaflops) computers before the end of the decade. Interestingly, computer scientists aren’t too concerned about storing or processing such huge amounts of data, but they are concerned about how humans will be able to make sense of it. It turns out, our brains are only capable of receiving information at rates in the gigabit range, so this much data would literally overwhelm us.

But computer scientists in Japan have proposed a creative way to compress the data, taking their cue from a filming technique known as “bullet time” (ala The Matrix). They’ve developed a visualization technique known as p-CAVE, a multi-modal system for analyzing and understanding output data from large-scale computer simulations. Like bullet time, it slows down the information enough to show events that are normally imperceptible and surrounds the “action” with thousands (or millions) of virtual cameras that record it from every angle. Users can then view the slow-motion data from a variety of perspectives by switching from one angle to the next.

The data compression is quite impressive. In a 10-gigabyte simulation of seismic waves recorded from 130 virtual cameras, the resulting movie ended up being 1.7 gigabytes. They expect the compression ratio to increase with larger scale simulations.