September 19, 2008 Having crashed through the petaflop barrier of a thousand trillion calculations per second back in June, scientists are continuing work into achieving the next benchmark in supercomputer performance - exascale computing. Georgia Institute of Technology's Professor Karsten Schwan is one of two scientists working in the field to receive a 2008 HP Labs Innovation Research Award to further research into solving the problems thrown-up by computing on this scale.
Exa is the prefix for quintillion, or 10^18, or a million trillion. That's a lot of calculations per second. Sustained performance of 1.02 petaflops was achieved by a system built by IBM for the Los Alamos National Laboratory (dubbed Roadrunner) a few months back. This put the machine well ahead of Blue Gene/L (with a performance of 478.2 teraflop/s) which comes in at number two on the list of top supercomputer sites.
In looking to go beyond these formidable numbers, researchers will have to develop new ways to build and run these monstrous machines.
Roadrunner is built on "hybrid" computing architecture that combines different types of processors, an approach also taken by Schwan.
“We believe that machines will reach exascale size only by combining common chips – such as quad core processors – with special purpose chips – such as graphics accelerators,” said Schwan, who is also director of the Georgia Tech Center for Experimental Research in Computer Systems (CERCS).
Schwan will undertake further research into how to efficiently run programs on these heterogeneous many-core platforms along with will Georgia Tech School of Electrical and Computer Engineering professor Sudhakar Yalamanchili.
Schwan will also collaborate with research scientist Ada Gavrilovska on the problem of managing multiple applications on exascale machines.
But why?
“The need for exascale-sized machines is well-established,” said Schwan. “With exascale machines, weather simulations will be able to operate at finer resolution, biologists will be able to model more complex systems, and businesses will be able to run and manage many applications at the same time on a single large machine.”
Dr. Stephane Lafortune, University of Michigan also received a collaborative research grant from HPLabs in the field of Exascale Datacenters. Further info and a full list of the 41 Research Award winners is available here.
