BATON ROUGE --- Currently, the world’s fastest supercomputers are at the petascale level, meaning they are capable of running 1,000 trillion calculations per second. But, what will happen when supercomputers move from petascale to exascale and become even faster, capable of running a million trillion calculations per second? What architecture and interfaces will the research community need to use these next-generation machines effectively?
A new research group comprised of scientists and engineers from LSU, University of Southern California Information Sciences Institute, University of Illinois Urbana-Champaign, University of Delaware and Sandia National Laboratories is addressing those questions to prepare scientific research for exascale supercomputers.
The National Science Foundation, or NSF, has funded this group, called the Exascale Point Design Study, to have a series of collaborative meetings throughout the next year to determine what needs to happen to develop large-scale computing systems.
NSF selected members of this group based on their accomplishments and expertise in various areas of computational science. Together, the group will discuss programming, hardware, applications, systems design and other challenges researchers will need to overcome to use exascale machines effectively.
LSU will host the first of these meetings on Thursday, March 19 and Friday, March 20, on campus. The lead researcher from LSU on this study is Thomas Sterling, a professor in the LSU Department of Computer Science. Sterling is a former NASA and Cal Tech scientist who invented the Beowulf computing cluster. At LSU, he leads a research group within the LSU Center for Computation & Technology, or CCT, working on the ParalleX project to develop parallel computing systems and architecture for large-scale supercomputers.
At the end of this series of meetings, the Exascale Point Design Study group will produce a report with their conclusions for NSF, which will serve as a prototype to build and design exascale machines for scientific research.
“We are about to enter a new era of scientific computing, and it is an exciting and rewarding challenge to work with distinguished researchers from institutions around the country to determine what the next supercomputers need to be,” Sterling said.
A new research group comprised of scientists and engineers from LSU, University of Southern California Information Sciences Institute, University of Illinois Urbana-Champaign, University of Delaware and Sandia National Laboratories is addressing those questions to prepare scientific research for exascale supercomputers.
The National Science Foundation, or NSF, has funded this group, called the Exascale Point Design Study, to have a series of collaborative meetings throughout the next year to determine what needs to happen to develop large-scale computing systems.
NSF selected members of this group based on their accomplishments and expertise in various areas of computational science. Together, the group will discuss programming, hardware, applications, systems design and other challenges researchers will need to overcome to use exascale machines effectively.
LSU will host the first of these meetings on Thursday, March 19 and Friday, March 20, on campus. The lead researcher from LSU on this study is Thomas Sterling, a professor in the LSU Department of Computer Science. Sterling is a former NASA and Cal Tech scientist who invented the Beowulf computing cluster. At LSU, he leads a research group within the LSU Center for Computation & Technology, or CCT, working on the ParalleX project to develop parallel computing systems and architecture for large-scale supercomputers.
At the end of this series of meetings, the Exascale Point Design Study group will produce a report with their conclusions for NSF, which will serve as a prototype to build and design exascale machines for scientific research.
“We are about to enter a new era of scientific computing, and it is an exciting and rewarding challenge to work with distinguished researchers from institutions around the country to determine what the next supercomputers need to be,” Sterling said.
Publish Date:
03-17-2009
