Nvidia announced Tuesday that its Tesla GPUs will power Titan, Oak Ridge National Lab’s newest supercomputer.
The computer, anticipated to be the fastest and most energy-efficient in the world, will be leveraged by the U.S. Department of Energy for medical and geoscience research, including climate change patterns.
Titan is expected to be at least twice as fast and three times more energy-efficient than today’s top-ranking computer based in Japan. These high expectations stem primarily from the multi-processing GPUs (rather than CPUs) on which the supercomputer will run.
According to Steve Scott, chief technology officer of Tesla products at Nvidia, the power of Titan will reaffirm the energy-saving potential of GPU-based computing in general. “GPUs are the right path, moving forward, given the power constraints we’re all facing,” Scott told CRN. “And seeing an organization [Oak Ridge National Lab] of this status, with such an important mission, is a huge validation of this direction.”
The lab’s GPU-powered machine will be built in phased deployments and expand upon the functionality of their current supercomputer, Jaguar. In 2012, Titan will begin stage two of deployment, incorporating Nvidia’s next-generation GPU architecture, code-named “Kepler.”
As the energy- and time-preserving capabilities of GPUs become more evident, so, presumably, will their adoption. While Titan will earn the title of “biggest machine ever based on GPUs,” Scott said he is confident it won’t be the last. As the efficiency of GPU-powered machines means reduced thermal output and cooling costs – a perk especially attractive to larger healthcare and financial organizations or universities fostering large data centers – the trend is expected to grow over the next few years.
“We need to find new ways of computing that are much more energy-efficient,” Scott said, “and performance comes primarily from power efficiency.”