Parallel Programming on Shared Address Space Multiprocessors.

Jaswinder Pal Singh
Computer Science Department at Princeton University

Abstract:

This tutorial will begin by discussing the process of developing high-performance parallel programs for parallel systems that support a shared address space programming model. It will use case studies of real applications to illustrate the techniques. After discussing the issues for small-scale symmetric multiprocessors and moderate-scale distributed shared memory systems (which increasingly comprise the set of tightly-coupled multiprocessors built in industry), it will examine the issues in programming for scalability to large processor counts on tightly-coupled systems and for performance portability to the same programming model supported in software on less tightly-coupled commodity clusters. The latter constitute the other dominant platform for scalable computing, with very different characteristics, and it is important to the future of a programming model that it be supported well on such platforms as well. Performance results on real systems will be provided to illustrate the issues throughout. The necessary parallel architecture and systems background will be introduced during the tutorial as necessary.

Bio:

Jaswinder Pal Singh is an Associate Professor in the Computer Science Department at Princeton University. He obtained his Ph.D. from Stanford University in 1993, and his B.S.E. degree from Princeton in 1987. His research interests are at the boundary of parallel and distributed applications and multiprocessor systems, both architecture and software, and in applications of high-performance computing.

At Stanford, he participated in the DASH and FLASH multiprocessor projects, leading the applications efforts there. He has led the development and distribution of the SPLASH and SPLASH-2 suites of parallel programs, which are widely used in parallel systems research. At Princeton, he currently leads the PRISM research group which does application-driven research in supporting programming models on a variety of communication architectures, as well as in novel applications of high-performance computing such as simulating the immune system. He has co-authored a graduate textbook called "Parallel Computer Architecture: A Hardware-Software approach." Singh is a recipient of a Sloan Research Fellowship and the Presidential Early Career Award for Scientists and Engineers (PECASE).