Shell Distinguished Lectures1998-1999 AbstractsA General Purpose Shared-Memory Model for Parallel ComputationVijaya Ramachandran,University of Texas at Austin 4:00pm, Thursday July 17, 1997
AbstractA fundamental challenge in parallel processing is to develop effective models for parallel computation that balance simplicity, accuracy, and broad applicability. In particular, a simple ``bridging'' model, i.e., a model that spans the range from algorithm design to architecture to hardware, is an especially desirable one. We propose the Queuing Shared Memory (QSM) model as a bridging shared-memory model for parallel computation. The QSM provides a high-level shared-memory abstraction for parallel algorithm design, as well as the capability to model bandwidth limitations and other features of current parallel machines, as evidenced by a randomized work-preserv ing emulation of the QSM on the BSP, which is a lower-level, distributed-memory model.
BiographyVijaya Ramachandran received her Ph.D. in EE and CS from Princeton Univ. in 1983. In Sept. 1995 she earned the title, Blakemore Regents Professor. Some recent publications: "An Optimal EREW PRAM Algorithm for Minimum Spanning Tree Verification," "An Efficient Parallel Algorithm for the Layered Planar Monotone Circuit Value Problem," "Parallel Algorithms for Reducible Flow Graphs," and "Parallel Implementation of Algorithms for Finding Connected Components in Graphs."3DDI: 3D Direct InteractionJohn Canny,University of California at Berkeley 4:00pm, Wednesday October 15, 1997
Abstract3DDI is about direct interaction with a simulated or remote 3D world. Users interact with the world without gloves or motion capture sensors and view the world stereoscopically without glasses. That is, they see objects which appear to exist in space in front of them, in full 3D. 3D interaction preserves the spatial relationships and body language cues among a group of people in ways that 2D video cannot. It allows people to interact with virtual devices as they would with their physical counterparts.The project covers the pipeline of technologies from real-time 3d capture devices, through physical modeling, to rendering on autostereoscopic and volumetric displays. This talk will summarize the project, and dive into detail on its physical simulation component. We will describe our simulator IMPULSE and its recent extensions to run *physical* behaviors encapsulated in VRML and Java. I will also talk briefly about another project on collaboration using "PRoPs". PRoPs are minimalist robotic avatars that roam the real world.
BiographyJohn Canny is a professor in the Computer Science Division at the University of California at Berkeley, which he joined in August 1987. He received his Ph.D. degree in 1987 from MIT for a thesis on Robot Motion Planning. His main research interests are human-computer interaction through computer graphics and robotics. This includes work in gestural input and physically-based simulation. He also works on manufacturing systems and novel manipulation methods down to micro-scale.What C++ Is and WhyBjarne Stroustrup,AT&T Research 4:00pm, Monday February 9, 1998
AbstractThis talk consists of three parts. First, the origins, aims, and design rules for C++ are described. Next, the key language features and the design and programming techniques they support are presented. Finally, I say a few words about current uses of C++, the C++ standardization effort, and likely future uses of C++.The emphasis is on the reasons why C++ looks the way it does and how it can be used well, rather than on syntactic details or individual language features. I will attempt to illustrate how C++ is the result of a combination of sociological and language-technical principles. The language features mentioned are the ones that most directly supports data abstraction, object-oriented programming, and the design and use of large systems: classes, abstract classes, templates, class hieararchies, run-time type information, namespaces and templates. Some examples will illustrate the standard library. This talk does not presuppose understanding of C++.
BiographyBjarne Stroustrup was born in Aarhus Denmark 1950. He received his Cand. Scient. Degree in Mathematics and Computer Science in 1975 from the University of Aarhus, Denmark, and his Ph.D. in Computer Science in 1979, from Cambridge University, England.Bjarne Stroustrup is the designer and original implementor of C++ and the author of ``The C++ Programming Language'' ( 1st edition 1985, 2nd edition 1991 3rd edition 1997) and The Design and Evolution of C++. His research interests include distributed systems, operating systems, simulation, design, and programming. Dr. Stroustrup is the head of AT&T Lab's Large-scale Programming Research department, an AT&T Bell Laboratories Fellow, and an AT&T Fellow. He is actively involved in the ANSI/ISO standardization of C++. He is a fellow of the ACM and the recipient of the 1993 ACM Grace Murray Hopper award. His non-research interests include general history, light literature, photography, hiking and running, travel, and music. He lives in Watchung, NJ with his wife, daughter, and son.
Scaling up to Very Large Parallel SystemsGreg Astfalk,Hewlett-Packard Company 4:00pm, Monday March 9, 1998
AbstractThere is a desire for dramatically increased levels of performance from computer systems. The applications to which this increased performance could be applied are quite diverse. Owing to this the systems that would be most useful are general-purpose. This leads us to the question of what would be the architecture of a general-purpose system of enormous computing power. As a target we examine what a system of greater than 5 Tflops would resemble within the next few years.Our examination looks at several possible ways of getting there (some of which are silly) and the technological challenges associated with each alternative. While it would be nice to maintain technological purity, we will cloud this talk with some genuine constraints. The dirty words we consider are; physics and economics. At the end of all of this we describe the impact on the users of the type of general-purpose system that we feel could actually reach the multi-Tflop level.
BiographyGreg Astfalk is the Chief Scientist at Hewlett-Packard's High Performance Systems Division (HPSD) in Richardson, TX. In this capacity he is involved in almost all the technologies of the systems that HPSD develops. This includes the processors, hardware, and software.Before assuming his current position he used to actually do scientific work, in the form of applied mathematics, for HP. He was with Convex Computer Corporation before it was acquired by Hewlett-Packard. Between Convex and HP he has 10 years in the high-end computing industry. Prior to Convex he was with AT&T Bell Labs doing numerical analysis for a very long time.
Smart Cooperative AgentsRuzena Bajcsy,University of Pennsylvania 4:00pm, Wednesday March 25, 1998
AbstractDuring the last six years we have been engaged in research regarding cooperative physical mobile agents that can observe and manipulate in an indoor environment. This problem has provided us with a varied and fruitful agenda of challenging scientific and engineering problems. Our recent accomplishments include:
The scientific challange is to design automatic switching mechanisms triggered by the environment and the task which select the "best" perceptual strategy to accomplish the task. This is, by the way, how biological systems function. Needless to say once we have good perception, the control of agents is easy. This talk describes collaborative work with Profs. Vijay Kumar, Jim Ostrowski, and Max Mintz of the GRASP Laboratory at the University of Pennsylvania.
BiographyRuzena Bajcsy obtained her first PhD in EE from Slovak Technical University (Czechoslovakia) in 1967 - she was the first woman in Slovakia ever to obtain a PhD. She then went to Stanford University and obtained her second PhD in 1972 studing Artificial Intelligence under John McCarthy. Her dissertation topic was machine perception, and she wrote one of the first programs enabling recognition of textured patterns.She then joined the faculty at the University of Pennsylvania where she has continued her work on machine perception and computer vision, characterizing and solving problems involving segmentation, 3-D vision, and other sensory modalities that function together with vision (e.g., touch). Professor Bajcsy's General Robotic Active Sensory Perception (GRASP) Laboratory at the University of Pennsylvania is internationally recognized in the field. She chaired her department from 1984 through 1986, and has served on numerous National Research Council and National Science Foundation advisory boards and committees. She was elected a fellow of IEEE in 1992 and of ACM in 1995; she is also a founding fellow of AAAI. In 1995 she was elected to membership in the National Institute of Medicine, and she became a member of the National Academy of Engineering in 1997. She has served for 3 years on the CRA Board.
The State of Programming Support Software for High Performance ComputersProf. Ken Kennedy,Rice University 4:00pm, Monday April 13, 1998
AbstractIn spite of the promise of scalable parallel computation and the substantive five-year Federal investment in the High Performance Computing and Communications Program, software technologies for HPC have fallen short of the goal of making parallel programming accessible to the non-specialist. This talk will survey the achievements and failures of software technologies for programming support, including compilers, run-time libraries, and tools for the construction tuning and debugging of programs for high-performance computer systems. It will then turn to an exploration of the challenges faced by developers of programming support software over the next decade, as high-end computer system performance approaches the petaops level and as applications move toward distributed, heterogenous computing platforms. The discussion will focus on the principal technology areas that will be increasingly important over the next decade:
BiographyKen Kennedy received a B.A. in mathematics from Rice University in 1967, an M.S. in mathematics from New York University in 1969, and a Ph.D. in computer science from New York University in 1971. Since 1971, he has been at Rice University where he founded the Department of Computer Science in 1984. Kennedy, holder of the Ann and John Doerr Professorship, is currently Director of the Center for Research on Parallel Computation, an NSF Science and Technology Center with seven prestigious participating institutions. In 1997 he was appointed co-chair of the Presidential Advisory Committee on High Performance Computing and Communications, Information Technology, and Next Generation Internet.His current research focuses on developing new strategies for supporting architecture-independent parallel programming. He has chaired the High Performance Fortran Forum, an informal standardization group that defined a set of extensions to Fortran 90 for data parallel programming. Professor Kennedy is a member of the National Academy of Engineering, a fellow of the AAAS and of the ACM and IEEE and the recipient of the 1995 W. W. McDowell Award, the highest research award of the IEEE Computer Society. |
| ||||
| ||||