|
|||||||||||||||||||||||||
|
|||||||||||||||||||||||||
 |
Simple, really |
||||||||||||||||||||||||
|
released 08.11.08 by J. William Bell A nationwide team of collaborators, embodied in the Great Lakes Consortium for Petascale Computation, will back the Blue Waters petascale computing project. Big changes mean teamwork. And Blue Waters is the definition of big—dramatically increasing the computational power available to America's researchers, fundamentally altering a host of supercomputing technologies, requiring new and enhanced software to make the most of those technologies, and obliging us to work with a wide variety of applications and codes from scientists to make sure that they scale to Blue Waters' unprecedented size. The Blue Waters petascale computing system will be a national asset, and it will have a nationwide team of collaborators backing it. The Great Lakes Consortium for Petascale Computation is a partnership among colleges and universities, national research laboratories, and other educational institutions. Led by NCSA and the University of Illinois, the consortium will be based on coordinated multi-institutional efforts. These projects will advance computational and computer science, engineering, technology research, and education. Early projects by members of the consortium will get underway immediately. Adolfy Hoisie and his team at Los Alamos National Laboratory, for example, will work on performance modeling of scientific applications on Blue Waters. Meanwhile, Louisiana State University's Center for Computation and Technology will help create a new generation of those applications, working with consortium partners and prospective users to build advanced tools that will make it easier for researchers to solve petascale science and engineering problems. The Cactus Computational Toolkit and the Eclipse open development platform will be at the center of their efforts. Cactus is used to create modular, collaborative high-performance computing applications that scale up to the largest systems available. Eclipse is focused on building software tools and libraries. The team will extend and combine these tools to help scientists and engineers take maximum advantage of Blue Waters and other petascale machines that will follow it. "Systems of Blue Waters' scale have never been built before, let alone used for the solution of complex problems. Advanced projects require not only petascale computing, networks, and data, but also groups of scientists, engineers, mathematicians, and computer scientists who can develop the complex software applications," says Ed Seidel, director of the Center for Computation and Technology. North Carolina's Renaissance Computing Institute will concentrate on performance monitoring and analysis on Blue Waters. "RENCI's role in Blue Waters will be to find the best ways to predict, monitor and evaluate performance and respond to problems before they become severe," said RENCI Interim Director Alan Blatecky. "It is critical to understand performance and reliability issues on any high-performance system, but especially on Blue Waters because there has never been a system of this scale before." Other related projects, including those led by consortium partners at Shodor and the University of Michigan, will focus on educating the next generation of researchers. Members will work together on new undergraduate and graduate courses that integrate petascale computing into disciplines other than computer science or engineering. Collaborators say the benefit and appeal of the entire enterprise is simple, really. The Great Lakes Consortium and the Blue Waters project allow them to partner with the world's best computational and discipline scientists, putting them squarely in the center of the scientific revolution being driven by advanced computing that will change the world. The Great Lakes Consortium for Petascale Computation includes:
|
||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||
Download a PDF of this story
Email this story
Subscribe to our newsletter
