Last modified: Wednesday, February 20, 2013
IU awarded $1.1 million to advance extreme scale computing
FOR IMMEDIATE RELEASE
February 20, 2013
BLOOMINGTON, Ind.— Indiana University's Center for Research in Extreme Scale Technologies (CREST) is the recipient of a three-year, $1.1 million grant from the Department of Energy (DOE) to develop software that improves the speed and programmability of supercomputers. This funding is part of a $7.05 million grant for the XPRESS (eXascale PRogramming Environment and System Software) project, led by Sandia National Laboratories as part of the DOE Office of Science Advanced Scientific Computing Research X-Stack program.
IU created CREST in 2011 as part of the Pervasive Technology Institute to pioneer research at the frontiers of exascale computing. Two of supercomputing's foremost thinkers, Andrew Lumsdaine and Thomas Sterling, both professors in the School of Informatics and Computing at IU Bloomington, lead CREST as director and associate director, respectively. Sterling also serves as CREST chief scientist.
The grant will fund CREST researchers to create a class of software that enables supercomputers to run intelligently. "We're writing software that moves execution from static to dynamic, allowing supercomputers to use new information as it is being revealed," said Sterling. "By doing so, supercomputers will 'think' about how they use their resources, as well as where and when they schedule various concurrent tasks."
As an analogy, Sterling noted the difference between a cannon and guided missile -- the missile makes minute adjustments during flight in order to more accurately hit the target. "Essentially, we're building a guided computer," said Sterling. "Our goal is to completely redesign the system software in order to produce a revolutionary class of supercomputers. It is exciting that IU will be at the forefront of such research, setting future directions for exascale computing and programming."
How fast is this next generation of supercomputers? Consider this: Today's fastest supercomputers perform about 10 quadrillion (one million billion) operations per second. By 2020, experts predict that exascale computers will perform one quintillion (one billion billion) calculations per second. However, it is not just speed that interests IU researchers—they are ultimately seeking to change how supercomputing works.
"This grant allows us to help scientists and engineers run their software across millions of processors," said Lumsdaine. "It's exciting to be able to advance next-generation supercomputing while supporting research into solutions for civilization's biggest issues."
Led by Sandia in partnership with IU, the XPRESS project involves eight academic and government institutions. Sterling is the project's chief scientist, while Lumsdaine is principal investigator on IU's portion of the grant.
About CREST
Indiana University's Center for Research in Extreme Scale Technologies (CREST) is the newest research center affiliated with the Pervasive Technology Institute. CREST's mission is to transform dynamic data-driven computing through the development and application of revolutionary high performance computing platforms. For more, see https://pti.iu.edu/crest/.
In 2008, IU established the Pervasive Technology Institute (PTI) through a $15 million grant from the Lilly Endowment. PTI is dedicated to the development and delivery of innovative information technology to advance research, education, industry and society. For more, see https://pti.iu.edu.
About extreme scale and exascale computing
Today's most powerful computers consist of over a million processor cores and deliver performance on the order of 10 petaFLOPS (equivalent to about a million laptop computers). Exascale computers will offer nearly 100 times that performance for real-world applications like climate modeling, microbiology, nuclear reactor design, combustion and mechanical deformation. Extreme scale computing not only includes exascale computing, but also delivers performance gains for problems that today cannot scale anywhere near the maximum system size available.
