ACADEMIA
Prof. Kale to Lead Project for Illinois Advanced Computing Institute
A project co-led by computer science professor Laxmikant Kale has been selected as one of the three inaugural projects for the new Institute for Advanced Computing Applications and Technologies at the University of Illinois at Urbana-Champaign. The Institute transfers advances from the computer science and engineering research at the National Center for Supercomputing Applications (NCSA) to the larger scientific, engineering, and arts, humanities and social science communities in order to speed progress across all of these frontiers. 
Kale will co-lead the Synergistic Research on Parallel Programming for Petascale Applications with Materials Science and Engineering professor Duane Johnson. Effectively harnessing the power of supercomputers (like the sustained-petascale Blue Waters system scheduled to come online at the University of Illinois in 2011) will require the coordinated development of petascale parallel programming tools and petascale applications. This project will combine potentially petascale applications- including codes for astrophysical (FLASH) and biomolecular (NAMD) simulation and for determining the electronic-structure of materials (QMCpack and MECCA)-with needed computer science research. Efforts will focus on adaptive runtime systems that automate dynamic load balancing and fault tolerance, enhancement of parallel programming abstractions, best-practice software engineering to petascale applications via refactoring tools, productive programming environments that integrate performance analysis and debugging tools, and automatically tuned libraries.
"Our major goal," said Johnson, "is to focus on applications having impact on challenging physical problems of broad community interest and that could really show sustained petascale performance given the right computer science tools and libraries on the planned hardware. Concerted effort between the physical and computer scientists is critical for this to happen-not all square blocks fit into the same round hole. The outcome is solutions to problems that, up till now, were impossible."
"These [inaugural] projects will bring the development and deployment competencies of NCSA to bear on challenges in diverse disciplines and will forge unique collaborations between Illinois faculty and NCSA staff," said Institute and NCSA Director Thom Dunning. "It's very exciting to be able to foster such innovative work."
The Institute is organized around five broad themes: Advanced Information Systems, Computing and Creativity, Data-intensive Applications and Technologies, Simulation of Natural and Engineered Systems, and the Center for Petascale Computing. Two of the initial projects are under the Simulation of Natural and Engineered Systems, while the third falls under Computing and Creativity.
For more information about the Institute, see www.iacat.uiuc.edu.
Kale will co-lead the Synergistic Research on Parallel Programming for Petascale Applications with Materials Science and Engineering professor Duane Johnson. Effectively harnessing the power of supercomputers (like the sustained-petascale Blue Waters system scheduled to come online at the University of Illinois in 2011) will require the coordinated development of petascale parallel programming tools and petascale applications. This project will combine potentially petascale applications- including codes for astrophysical (FLASH) and biomolecular (NAMD) simulation and for determining the electronic-structure of materials (QMCpack and MECCA)-with needed computer science research. Efforts will focus on adaptive runtime systems that automate dynamic load balancing and fault tolerance, enhancement of parallel programming abstractions, best-practice software engineering to petascale applications via refactoring tools, productive programming environments that integrate performance analysis and debugging tools, and automatically tuned libraries.
"Our major goal," said Johnson, "is to focus on applications having impact on challenging physical problems of broad community interest and that could really show sustained petascale performance given the right computer science tools and libraries on the planned hardware. Concerted effort between the physical and computer scientists is critical for this to happen-not all square blocks fit into the same round hole. The outcome is solutions to problems that, up till now, were impossible."
