The worldwide scientific research community is anticipating the arrival of petascale computing, the next generation of supercomputers that will enable better research and allow breakthroughs that are not possible with current high-performance computing systems. Researchers with LSU’s Center for Computation & Technology, or CCT, are developing graphics processing units and software applications through the Cactus Computational Toolkit so that when these systems deploy, scientists will have the tools they need to advance discovery.
As an example of the power petascale machines would have, it would take approximately 11,000 home computers plus advanced networking and data storage connections to equal the current supercomputing power in Louisiana through LSU’s supercomputers and the Louisiana Optical Network Initiative machines. When the research community gains petascale capabilities, that will represent the processing power of 200,000-500,000 home computers.
Blue Waters: Breaking Through the Limits
LSU is part of the Great Lakes Consortium for Petascale Computation, which will play an integral role in Blue Waters, the first full petascale supercomputer. The Blue Waters project is headquartered at the University of Illinois Urbana-Champaign’s National Center for Supercomputing Applications, or NCSA, which is a collaborative partner with the CCT on many endeavors. The Blue Waters project has received the recommendation of the National Science Board and has signed a cooperative agreement with the National Science Foundation.
CCT and partners at NCSA and NVIDIA Corporation are working on a project to upgrade graphics processor units, or GPUs, from the NVIDIA Tesla series to make these GPUs compatible for petascale computing. Researchers will need massive, fine-grained parallel computing capabilities to effectively use petascale computing environments.
The Institute for Advanced Computing Applications and Technologies, or IACAT, at the University of Illinois at Urbana-Champaign recently launched a project to empower science and engineering researchers by enabling their applications to run 100 times faster and at much lower cost than on traditional parallel processing techniques. CCT and NVIDIA are assisting with this project to develop models, tools and applications communities that will leverage the use of GPUs for petascale computing.
Researchers from both universities will develop application communities in key engineering and scientific areas, including computational biophysics, astrophysics and computational fluid dynamics, along with emerging disciplines that rely on high-performance computing, such as digital arts and humanities. This will provide scenarios for researchers to test the developing applications and tools.
Researchers in the CCT’s Cactus Computational Toolkit group are also developing the new algorithms, tools and visualization capabilities needed for petascale computing and beyond. 
Since petascale computing will allow scientists to make complex, high-resolution models of physical processes including climate change, hurricane behavior, oil extraction from reservoirs and astrophysical black hole collisions and gamma ray bursts, they will these models will need to run effectively on petascale computing architectures so scientists can better understand and use them. This involves leveraging many-core processors, hardware accelerators and new programming paradigms --- scaling to hundreds of thousands of processors. The Cactus Framework already provides an advanced parallel programming environment and community toolkits for collaborative problem solving, so LSU researchers are developing new applications to make it run models effectively on petascale machines.
Cactus is funded by National Science Foundation, Department of Energy, NASA and the Department of Defense. Researchers developed Cactus at LSU and the Albert-Einstein-Institute in Berlin.
For more information on CCT’s research projects on petascale computing, please visit www.cct.lsu.edu or contact CCT Manager of Public Relations Kristen Sunde at 225-578-3469.
For more information on progress in petascale computing, please visit:
Publish Date:
04-30-2008
