Spring 2010 Distinguished Lecture AbstractsComputer Science and Engineering Distinguished Lecturer SeriesThe new era in genomics: Opportunities and challenges for high performance computing
Dr. Srinivas Aluru
4:10 p.m., Monday, February 15, 2010 AbstractFor nearly three decades, the Sanger method of reading one DNA sequence at a time served as the mainstay of genomics research. The advent of high-throughput short read DNA sequencing technology is enabling revolutionary advances in life sciences by providing an inexpensive way to sample genomes at high coverage. This is causing a paradigm shift in individual genome sequencing, sequencing unknown genomes and metagenomes, and generating transcript expression profiles. The rate of data generation is handily outpacing Moore's law, with tenfold increase per year. This talk will outline the computational challenges in exploiting high-throughput sequencing technology, and its potential applications. I will focus on the genome sequencing and assembly problem, and illustrate the role of parallel algorithmic methods as we transition to the era of high-throughput sequencing. BiographySrinivas Aluru is the Mehl Professor of Computer Engineering at Iowa State University, and the Bajaj Chair Professor of Computer Science and Engineering at Indian Institute of Technology Bombay. Earlier, he served as Chair of Iowa State's Bioinformatics and Computational Biology program, and held faculty positions at New Mexico State University and Syracuse University. Aluru conducts research in high performance computing, bioinformatics and systems biology, combinatorial scientific computing, and applied algorithms. He pioneered the development of parallel methods in computational biology, and contributed to the sequencing and analysis of the maize genome. He is a Fellow of the IEEE, and is a recipient of the NSF Career award, the Swarnajayanti fellowship from the Government of India, IBM faculty award, and Iowa State University Foundation award for mid-career achievement in research. Faculty Contact: Dr. Tiffani Williams (tlw [at] cse.tamu.edu) ECE Leaders and Innovators SeriesMulti-core, Many-core, Mega-nonsense
Dr. Yale N. Patt
3:55 p.m., Tuesday, March 30, 2010 AbstractProcess technology is promising 50 billion transistors on each silicon die within the next few years, and the hype that this has generated is almost overwhelming. In this talk, I hope to examine a dozen of the claims we hear for multi-core and discuss what is real and what is abject nonsense. Such things as ILP is dead, hardware is sequential, parallel programming is hard, simple homogeneous cores make sense, Moore's Law is now all about doubling the number of cores on the chip - these form the tip of the iceberg. The hope is that by examining the nonsense, we can find a way to move forward. BiographyYale Patt is a teacher at The University of Texas at Austin, where he enjoys equally teaching the required intro to computing course to 400+ freshmen, and the advanced graduate course in microarchitecture to prepare students for PhD research and/or positions on chip design teams in industry. Years ago he earned the appropriate degrees from reputable universities, and more recently has received more than enough awards for his research and teaching. For those who are interested, details can be found on his website at http://www.ece.utexas.edu/~patt. Faculty Contact: Dr. Narasimha Reddy (reddy [at] ece.tamu.edu) Computer Science and Engineering Distinguished Lecturer SeriesHigh Performance Microprocessors ten years from now: What are the problems? How do we solve them?
Dr. Yale N. Patt
4:10 p.m., Wednesday, March 31, 2010 AbstractProcess technology is promising 50 billion transistors on each silicon die within the next few years. Our job - how to harness them to keep on the performance curve we have enjoyed for the last few decades. Everyone knows there are issues with power and energy. Some recognize that there are also new bandwidth and old latency issues. What else? In this talk, I hope to discuss some of the issues and suggest what the microprocessor of the future may look like. I will also indicate the role education must play if we are to be successful. BiographyYale Patt is a teacher at The University of Texas at Austin, where he enjoys equally teaching the required intro to computing course to 400+ freshmen, and the advanced graduate course in microarchitecture to prepare students for PhD research and/or positions on chip design teams in industry. Years ago he earned the appropriate degrees from reputable universities, and more recently has received more than enough awards for his research and teaching. For those who are interested, details can be found on his website at http://www.ece.utexas.edu/~patt. Faculty Contact: Dr. Lawrence Rauchwerger (rwerger [at] cse.tamu.edu) Computer Science and Engineering Distinguished Lecturer SeriesThe Challenge of the Multicores
Fran Allen
4:10 p.m., Wednesday, April 7, 2010 AbstractComputers have traditionally used complex "cores" to manage and execute computations and, over the years, these computers have achieved ever increasing levels of performance. However fundamental limitations of size, heat and energy have emerged. To overcome these limitations complex cores are being displaced by multicores that are not only simpler and more numerous but can be used more flexibly. So what is the challenge? Multicore computers are ushering in a new era of parallelism everywhere. As more cores are available, the potential performance of the system can increase at the traditional rate by the use of the inherent parallelism. Or can it? How will users and applications take advantage of all the parallelism? This talk will review some of the history of languages and compilers for high performance systems and consider opportunities for performance on multicore systems. The talk is intended to encourage the exploration of new approaches including how users will continue to code sequential for parallel systems. BiographyFran Allen's specialty is compilers and program optimization for high performance computers. Soon after joining IBM Research as a programmer in 1957 with a University of Michigan masters degree in mathematics, Fran found the technical goal that would drive her career: Enable both programmer productivity and program performance in the development of computer applications. One outcome of her work is that Fran received ACM's 2006 Turing Award "For pioneering contributions to the theory and practice of optimizing compiler techniques that laid the foundation for modern optimizing compilers and automatic parallel execution." Fran is a member of the American Philosophical Society and the National Academy of Engineers, and is a Fellow of the American Academy of Arts and Sciences, ACM, IEEE, and the Computer History Museum. Fran holds several honorary doctorate degrees and has served on numerous national technology boards. Fran is an active mentor, an advocate for technical women in computing, an environmentalist and an explorer. Faculty Contact: Dr. Nancy Amato (amato [at] cse.tamu.edu) Fall 2009 Distinguished Lecture AbstractsComputer Science and Engineering Distinguished Lecturer SeriesComputing the Entropy of Two-dimensional Shifts of Finite Type
Dr. Brian Marcus,
4:10 p.m., Wednesday November 11, 2009 AbstractA one-dimensional shift of finite type (SFT) is the set of infinite sequences that do not contain, as a sub-word, any finite word in a given finite list. The simplest example is the golden mean shift, which is defined as the set of all infinite binary sequences which do not contain the word "11" (i.e.., 1's are isolated). SFT's are ubiquitous as models of dynamical systems and also as constraints imposed on sequences to improve the performance of data recording systems. Perhaps the most fundamental quantity associated to an SFT is its entropy (called "topological entropy" in dynamical systems and "capacity" in information theory); entropy is defined as the asymptotic growth rate of the number of allowed finite words in the system. The entropy is easily computable as the log of the largest eigenvalue of a nonnegative integer matrix. For instance, the entropy of the golden mean shift is the log of the golden mean. A two-dimensional SFT is defined as the set of tilings of the integer lattice that do not contain as a sub-array any finite array in a given finite list. Two-dimensional SFT's are much less understood than their one-dimensional counterparts. In particular, there is no known closed-form expression for the entropy, which is defined as the asymptotic growth rate of the number of allowed arrays in the system. Even for the simple case of the two-dimensional golden mean shift (also known as the hard square model), which is defined as the set of all binary tilings that do not contain two adjacent 1's, horizontally or vertically, there is no known explicit formula. In this talk, we present recent results in joint work with Erez Louidor and Ronnie Pavlov. These include improved numerical approximations to entropy of specific SFT's and their cousins (sofic shifts), numerical approximation schemes which are provably exponentially accurate for a class of SFT's including the two-dimensional golden mean shift, and a few new exact computations of entropy. BiographyBrian Marcus received his BA from Pomona College and PhD from UC Berkeley, both in Mathematics. His main research interests are in symbolic dynamics and information theory. He has been on the faculty of the University of North Carolina at Chapel Hill and the Research Staff of the IBM Almaden Research Center. He has held visiting and adjunct positions at several universities, including UC Berkeley and Stanford University. He has been Professor of Mathematics at the University of British Columbia since 2002, serving as Head of the Mathematics Department from 2002 to 2007. He is an IEEE Fellow, was co-recipient of the IEEE Leonard G. Abraham Prize Paper Award, co-author of "An Introduction to Symbolic Dynamics and Coding" (Cambridge University Press), has published extensively in Mathematics and Engineering journals and holds twelve US patents. Faculty Contact: Dr. Anxiao (Andrew) Jiang (ajiang [at] cse.tamu.edu) |
| ||||
| ||||