Nvidia GPUs to Power the World's Top Supercomputer

By Beecher Tuttle October 11, 2011

The U.S. Department of Energy's Oak Ridge National Laboratory will use Nvidia chips to power what it hopes will be the fastest and most powerful supercomputer in the world.

The new system, dubbed Titan, should be completed by 2012, putting the U.S. back on pace with Asian countries like China and Japan, which own the top 5 supercomputers in the world, according to the Wall Street Journal. Supercomputers are generally used for research in the fields of material science, energy technology, medical research and geoscience, among others.

Unlike the U.S.'s legacy supercomputers, Titan will be loaded with graphics processing units (GPUs). The Nvidia Tesla GPUs will help enable Titan to deliver over 20 petaflops of peak performance, making it more than two times faster and three times more energy efficient than today's fastest supercomputer, according to the chip maker.

Nvidia also built GPUs for China's "Tianhe-1A" supercomputer, which is currently ranked second on the list of the 500 fastest supercomputers.

Steve Scott, chief technology officer of Tesla products at NVIDIA, said that Oak Ridge's decision to embrace GPU technology for supercomputers – along with similar moves from China and Japan – underscores "the growing belief that GPU-based heterogeneous computing is the best approach to reach exascale computing levels within the next decade."

Scott said that Tesla GPUs will provide over 85 percent of the peak performance of Titan.

Jeff Nichols, associate laboratory director for Computing and Computational Sciences at Oak Ridge, said that the new supercomputer will be used to help develop "more commercially viable biofuels, cleaner burning engines, safer nuclear energy, and more efficient solar power."

The first step in the transition, currently underway, involves upgrading Oak Ridge's current Jaguar supercomputer with 960 Tesla M2090 GPUs based on the Nvidia's "Fermi" architecture. In 2012, the lab will leverage a next-generation Nvidia architecture named "Kepler" to deploy around 18,000 Tesla GPUs, taking Titan to what it hopes will be the top spot among supercomputers.


Beecher Tuttle is a TechZone360 contributor. He has extensive experience writing and editing for print publications and online news websites. He has specialized in a variety of industries, including health care technology, politics and education. To read more of his articles, please visit his columnist page.

Edited by Rich Steeves

TechZone360 Contributor

SHARE THIS ARTICLE
Related Articles

Apple, Cisco Join Forces - What It Could Mean for Them, and the Rest of Us

By: Paula Bernier    9/2/2015

Tech powerhouses Apple and Cisco systems this week announced they have joined forces to give business users of iOS devices the best possible experienc…

Read More

LTE Broadcast, IndyCar Racing, and Blending the Best of TV and Stadium Viewing

By: Rob Enderle    9/1/2015

This last weekend I was at the IndyCar race in Sonoma to see Verizon and Qualcomm showcase LTE Broadcast-I also wanted to see the race. I've watched N…

Read More

Windows 95 to Windows 10: How Far We Have Come in 20 Years

By: Rob Enderle    9/1/2015

Last week was the anniversary of Windows 95, which was actually a life-changing launch for me. I'd just started out as an analyst, and Windows 95 was …

Read More

Microsoft Research Project Allows for Inexpensive 3D Scanning from a Smartphone

By: Christopher Mohr    8/27/2015

It is now possible to perform 3D scanning from a smartphone, without additional hardware or an Internet connection, thanks to a new Microsoft Research…

Read More

Amazon's Scaled Back Consumer Device Efforts, Dash Button, and More

By: Paula Bernier    8/27/2015

Word is that Amazon is scaling way back on its consumer devices efforts, having let go of dozens of Lab126 engineers who worked on its Fire phone, acc…

Read More