[Slides]
Abstract
Serial computing has become largely irrelevant for growth in computing performance at around 2003. All processor vendors have presented many-core roadmaps expecting exponential increase in the number of cores over at least a decade. The whole world of computing is now facing the challenge of coming up with a truly general-purpose parallel computing platform-the same challenge that eluded high performance computing (HPC) for so many years--and the clock is ticking. It is becoming common knowledge that if you want your program to run faster you will have to program for parallelism, but the vendors who set up the rules have not yet provided clear and effective means (e.g., programming models and languages) for doing that.
The PRAM-On-Chip project started at UMD in 1997 foreseeing this challenge and opportunity. Building on PRAM - a parallel algorithmic approach that has never been seriously challenged on ease of thinking, or wealth of its knowledge-base - a comprehensive and coherent platform for on-chip general-purpose parallel computing has been developed and prototyped. The approach goes after any type of application parallelism regardless of its amount, regularity, or grain size. Some prototyping highlights include: an eXplicit Multi-Threaded (XMT) architecture, a new 64-processor, 75MHz XMT (FPGA-based) computer, 90nm ASIC tapeout of the key interconnection network component, a basic compiler, class tested programming methodology where (even high-school) students are taught only parallel algorithms and pick the rest on their own, and up to 100X speedups on applications.
The talk will give special attention to how this material was presented to a dozen high school students (10 from Montgomery Blair, 1 from Walter Johnson and 1 from Thomas Jefferson) in Fall 2007: a full day tutorial was followed up with only a weekly office-hour by an undergraduate assistant. Some strong students have been able to complete 5 of 6 assignments given in a graduate course on parallel algorithms. This informal course was part of a computer club that met after 8 hours of regular classes (i.e., no academic credit). This suggests that high school students from diverse technical backgrounds and abilities, hopefully everybody who can study standard serial programming, should be able to study this material if taught in a regular for-credit class by a professional teacher.
URL: http://www.umiacs.umd.edu/users/vishkin/XMT
About the Speaker
Uzi Vishkin got his DSc degree from the Technion, Israel in 1981. He has been a Professor of Electrical and Computer Engineering at the University of Maryland Institute for Advanced Computer Studies (UMIACS) since 1988. His work-depth methodology for presenting parallel algorithms provided the presentation framework in several parallel algorithm texts that also include many of his parallel algorithms. He is the inventor of the PRAM-On-Chip desktop supercomputer framework under development since 1997 at UMD. He was elected ACM Fellow in 1996 for, among other things, having "played a leading role in forming and shaping what thinking in parallel has come to mean in the fundamental theory of Computer Science", is an ISI-Thompson Highly Cited Researcher was recently named a Maryland 2007 Innovator of the Year for his PRAM-On-Chip venture.
