Oak Ridge National Laboratory will tap Nvidia chips to power what is expected to be one of the world's fastest supercomputers.
Oak Ridge's Titan supercomputer will eventually pack as many as 18,000 Nvidia graphics processing units (GPUs) and have the potential to deliver 20 petaflops of peak performance, making it one of the fastest computers in the world.
Last year, Nvidia made a splash when it announced that its chips were powering the Chinese "Tianhe-1A" supercomputer, which, at that time, became the fastest in the world. As of June, the Chinese system was ranked No. 2 in the world on the Top500.org Web site. Oak Ridge's current design was ranked No.3.
High-end GPUs typically contain hundreds of small processing cores, allowing them to accelerate certain types of computational tasks more efficiently and thereby much faster than CPUs (central processing units).
Oak Ridge National Laboratory, which operates a computing facility for the U.S. Department of Energy, had been testing a small-scale system roughly modeled on the Chinese supercomputer, according to Jack Dongarra, a professor at University of Tennessee's Department of Electrical Engineering and Computer Science, who spoke to CNET last year. He is part of a group participating in the Oak Ridge project.
If the much larger scale Titan system hits its performance goal, it would be more than two times faster and three times more energy efficient than today's fastest supercomputer, the K computer located in Japan, according to Nvidia.
In the first phase of the Titan deployment, which is currently under way, Oak Ridge will upgrade its existing Jaguar supercomputer with 960 Tesla M2090 GPUs based on the Nvidia "Fermi" architecture. These GPUs will serve as companion processors to multi-core CPUs in this Cray XK6 supercomputer.
In the second phase, expected to begin in 2012, Oak Ridge plans to deploy up to 18,000 Tesla GPUs based on the next-generation Nvidia architecture code-named "Kepler," Nvidia said.
"Titan will be used for a variety of important research projects, including the development of more commercially viable biofuels, cleaner burning engines, safer nuclear energy, and more efficient solar power," said Jeff Nichols, associate laboratory director for Computing and Computational Sciences at Oak Ridge National Laboratory.