 |
Invited Talks
|
H.J. Siegel Colorado State University, USA |
Monday July 5th 9.00 AM |
|
The Robustness of Resource Allocation in Parallel and Distributed Computing
Systems Performing computing and communication tasks on parallel and distributed systems involves the coordinated use of different types of machines, networks, interfaces, and other resources. Decisions about how best to allocate resources are often based on estimated values of task and system parameters, due to uncertainties in the system environment. An important research problem is the development of resource management strategies that can guarantee a particular system performance given such uncertainties. We have designed a methodology for deriving the degree of robustness of a resource allocation - the maximum amount of collective uncertainty in system parameters within which a user-specified level of system performance (QoS) can be guaranteed. Our four-step procedure for deriving a robustness metric for an arbitrary system will be presented. We will illustrate this procedure and its usefulness by deriving robustness metrics for some example distributed systems. This research was performed with Prof. S. Ali, Prof. A. A. Maciejewski, and Mr. J-K Kim. Speaker H. J. Siegel is the Abell Endowed Chair Distinguished Professor of Electrical and Computer Engineering at Colorado State University (CSU), where he is also a Professor of Computer Science. He is Director of the CSU Information Science and Technology Center (ISTeC), a university-wide organization for promoting, facilitating, and enhancing CSU's research, education, and outreach activities pertaining to the design and innovative application of computer, communication, and information systems. From 1976 to 2001, he was a Professor in the School of Electrical and Computer Engineering at Purdue University. He received two B.S. degrees from the Massachusetts Institute of Technology (MIT), and the M.A., M.S.E., and Ph.D. degrees from Princeton University. He is a Fellow of the IEEE and a Fellow of the ACM. Prof. Siegel has co-authored over 300 published technical papers in the areas of parallel and distributed computing and communications. He was a Coeditor-in-Chief of the Journal of Parallel and Distributed Computing, and was on the Editorial Boards of the IEEE Transactions on Parallel and Distributed Systems and the IEEE Transactions on Computers. |
|
|
Jack Dongarra University of Tennessee, USA Oak Ridge National Laboratory, USA |
Monday July 5th 10.30 AM |
|
Self Adapting Numerical Software (SANS) Effort and Fault Tolerant
Computing Self Adapting Numerical Software (SANS) is a software effort that will automatically generate highly optimized numerical kernels for our high performance computers. As the underlying computing hardware doubles its speed every eighteen months, it often takes more than a year for software to be optimized or "tuned" for performance on a newly released CPU. Users tend to see only a fraction of the power available from any new processor until it is well on the way to obsolescence. We will look at how parallel numerical library software can be run in an adaptive fashion to take advantage of available resources. We will also look at an approach for developing fault tolerant numerical computations using MPI. Speaker Jack Dongarra received a Bachelor of Science in Mathematics from Chicago State University in 1972 and a Master of Science in Computer Science from the Illinois Institute of Technology in 1973. He received his Ph.D. in Applied Mathematics from the University of New Mexico in 1980. He worked at the Argonne National Laboratory until 1989, becoming a senior scientist. He now holds an appointment as University Distinguished Professor of Computer Science in the Computer Science Department at the University of Tennessee and is holds the title of Distinguished Research Staff in the Computer Science and Mathematics Division at Oak Ridge National Laboratory (ORNL). Jack Dongarra specializes in numerical algorithms in linear algebra, parallel computing, use of advanced-computer architectures, programming methodology, and tools for parallel computers. His research includes the development, testing and documentation of high quality mathematical software. He has contributed to the design and implementation of the following open source software packages and systems: EISPACK, LINPACK, the BLAS, LAPACK, ScaLAPACK, Netlib, PVM, MPI, NetSolve, Top500, ATLAS, and PAPI. He has published approximately 200 articles, papers, reports and technical memoranda and he is coauthor of several books. He is a Fellow of the AAAS, ACM, and the IEEE and a member of the National Academy of Engineering. |
|
|
Martin Rem Embedded Systems Institute, University of Eindhoven, Netherlands |
Tuesday July 6th 9.00 AM |
|
The Role of Software in Embedded Systems Embedded systems differ from mechanical or mechatronic systems in that they are software intensive. Actually, a number of technologies meet in embedded systems: always electronics and software, and usually mechanics and dynamics, and sometimes other technologies such as optics. These technologies are used to make systems that satisfy certain requirements with respect to system-level qualities such as reliability, speed, cost, interoperability, etc. The design of embedded systems is, consequently, a multidisciplinary effort that tries to bridge the distance between the monodisciplinary components and the multidisciplinary system requirements. Software plays at least two roles in these systems. Embedded systems contain software components, just like they contain electronic or mechanical components, but software is also used to integrate different components and subsystems. The latter type of software is the main cause why the reliability of systems tends to decrease when they contain more software. With reliability as an example, system-level questions are addressed and it is analyzed how such questions may be tackled. The talk is based on research that is carried out in the Embedded Systems Institute (ESI). The way of working ("industry-as-laboratory") and the structure ("network institute") of the ESI are briefly explained. Speaker Martin Rem (1946) is scientific director of the Embedded Systems Institute (ESI) in Eindhoven, Netherlands. The ESI is a recently established institute for industrial research in embedded systems. It is a public-private partnership, jointly founded by industrial companies and universities. Martin Rem holds a PhD from the Technical University Eindhoven. Professor Edsger W. Dijkstra was his advisor. The dissertation ("Associons and the Closure Statement") addresses a model for parallel computing. He has worked at the Computer Science Department of the California Institute of Technology in Pasadena until he became a professor at Eindhoven. From 1991 to 1994 he was Dean of the Department of Mathematics and Computer Science. From 1996 to 2001 he was Rector of the university. |
|
|
Edwin K. P. Chong Colorado State University, USA |
Wednesday July 7th 9.00 AM |
|
Dynamic Resource Management In this talk we focus on resource management problems that are in some sense inherently dynamic. As our main example, we show that a broad class of problems in the control and management of computer/communication networks are inherently dynamic in this sense. To tackle such problems, we describe an approach to designing resource management algorithms that incorporate information from dynamic load (traffic) models. The approach can be applied to a wide variety of resource management problems. We have constructed experimental implementations of our approach for several resource management algorithms, and shown for these examples that the approach dramatically improves the quality of the resulting management policy. Our basic approach is based on combining dynamic load models with Markov decision theory. Dynamic load models can be viewed as stochastic predictions about the future system state. These predictions can be used to generate traces of potential future load behavior. Our approach is to use such predictions to heuristically evaluate candidate control actions. The principle underlying this approach draws from recent approximation/sampling methods for solving large Markov decision problems. This talk will describe two such methods --- policy rollout and hindsight optimization --- and show how these methods can be applied to a variety of dynamic resource management problems. Speaker Edwin K. P. Chong is a Professor of Electrical and Computer Engineering, and Professor of Mathematics, at Colorado State University. He received the B.E.(Hons.) degree with First Class Honors from the University of Adelaide, South Australia; and the M.A. and Ph.D. degrees from Princeton University. He joined the School of Electrical and Computer Engineering at Purdue University in 1991, where he was named a University Faculty Scholar in 1999, and promoted to Full Professor in 2001. He spent a sabbatical at Bell Labs, Lucent Technologies, in 1998. He joined Colorado State University in 2001, where he is currently serving as Lead Scientist for Colorado State University's Colorado Grid Computing effort. His current interests are in computer/communication networks, resource management, and optimization methods. He coauthored the best-selling book, An Introduction to Optimization, 2nd Edition, Wiley-Interscience, 2001. He received the NSF CAREER Award in 1995 and the ASEE Frederick Emmons Terman Award in 1998. He coauthored a paper that was awarded Best Paper in the journal Computer Networks, 2003. He is a Fellow of the IEEE. He was founding chairman of the IEEE Control Systems Society Technical Committee on Discrete Event Systems, and, until recently, served as an IEEE Control Systems Society Distinguished Lecturer. He has been on the editorial board of the IEEE Transactions on Automatic Control. He is currently on the editorial board of the Journal of Computer Networks. | |
Copyright © University College Cork 2004