2008 is likely to be the year we achieve PetaFLOPs Computing — computation at a rate of 1015 floating-point operations per second. The high-end systems developed to meet this aggressive goal continue to scale up and become more complex, and the process of porting scientific applications to each new platform generation is quite difficult. In this talk I will focus on current research in the High Performance Computing Laboratory that aims to develop a next generation of performance measurement and analysis tools to help with the optimization and analysis of application codes on platforms such as Linux clusters, Cray supercomputers, or the most powerful system in the world, Blue Gene/L. In particular I will discuss PerfTrack, a tool for storing and analyzing performance data from different program runs and different measurement tools; Trace Profiling, a promising new approach for event tracing that scales to a high number of processors; and Environment Aware Performance Analysis, an innovative view of application performance tuning that considers effects of the runtime environment in addition to the application itself.
Dr. Karavanic earned her B.A. in Computer Science from New York University, and her M.S. and Ph.D. in computer science from the University of Wisconsin — Madison. She teaches courses in operating systems and performance analysis and directs the High Performance Computing Laboratory, where students conduct course projects and advanced research, and collaborate with a variety of scientists in industry and research laboratories across the U.S. Dr. Karavanic has served on the program committees of numerous international conferences, including the ACM International Conference on Supercomputing (ICS), the International Conference on High Performance Computing and Communications (HPCC), ACM Computing Frontiers, the International Workshop on High-Level Parallel Programming Models and Supportive Environments (HiPS), and Supercomputing; and is a member of the Euro-Par Conference Advisory Board.
Andrew Black