NA Digest Sunday, May 22, 1988 Volume 88 : Issue 21
Today's Editor: Cleve Moler
Today's Topics:
-------------------------------------------------------
From: Greg Astfalk <astfalk%c1east%[email protected]>
Date: Wed, 18 May 88 11:07:50 edt
Subject: 3rd Parallel Circus
The 3rd Parallel Circus was held at IBM in Kingston on May 6-7. There
were no clowns nor elephant dropings but lots of good discussions on
parallel processing. Approximately 50 people attended during the two
days of the Circus. The first day, chaired by Gene Golub, was devoted
to linear algebra. Martin Schultz chaired the second day sessions
which finished up the linear algebra talks and covered particle methods.
Gene has set-up a NAnet address for the Parallel Circus so that comments,
questions, etc. can be sent to [email protected].
Following are some comments, a list of the talks and list of attendees.
Any omissions or typos in names or addresses are unintentional; flames to
Greg Astfalk
Convex Computer Corp.
convex!c1east!astfalk
--- C O M M E N T S ---
Comments made by attendees during the course of the circus were
favorable and unanimously in favor of continuation of the circus in
the future. Some distinct advantages are the exposure to as yet
unpublished work or work that will not be seen for some time. As a
specific example Craig Douglas's presentation covered material that is
accepted for publication but will not appear until December.
The informal nature of the Circus did not at all detract from the
quality of the presentations. Future meetings should consider having
shorter talks combined with longer breaks. Having non-stop talks is
too 'intense' and precludes the ability to talk one-on-one to
colleagues, which in many cases is more informative than the talks
themselves.
The next Circus is tentativley being planned for this Fall at Rutgers
with a spring '89 meeting at RPI. Stay tuned to na-net for details.
Thanks are to be extended to the IBM Center for Scientific Engineering
Computations and Enrico Clementi for providing the facilities for the
meeting. A special thanks goes to Vijay Sonnand for tending to the
organizational details.
--- T A L K S ---
Ravi Mirchandaney, Yale Univ, "The architecture of the PARTY runtime
system"
David Keyes, Yale University, "Domain decomposition techniques for
nonsymmetric systems"
Joel Saltz, Yale University, "Runtime techniques for parallel
execution of sparse matrix computations"
Bo Kagstrom, University of Umea, "Parallel algorithms for solving the
triangular Sylvester equation on a Hypercube multiprocessor"
Robert van de Geijn, Univ of Texas at Austin, "A new storage scheme for
parallel implementation of eigenvalue algorithms"
Apostolos Gerasoulis, Rutgers, "Gaussian elimination and Gauss Jordan
on MIMD architectures"
Wei-Pai Tang, Univ. of Waterloo, "Wavefront elimination,
renormalization and fractals"
Stefano Foresti, IBM - Kingston, "Parallel multilevel iteration for a
p-version of finite elements"
Steve Hammond, "Solution of large, sparse, linear systems using ICCG
on a linear systolic array"
Gautam Shroff, RPI, "Convergence proofs for parallel block Jacobi
methods"
Christian Bischof, Cornell, "A pipelined QR algorithm with variable
blocking"
Anita Mayo, IBM, "Volume integrals and particle methods"
Josh Barnes, IAS, "Multiple-timestep Parallel Algorithms"
Enrico Clementi, IBM - Kingston, "Parallelism in physics, chemistry
and simple engineering"
Craig Douglas, IBM - Kingston, "Parallel multigrid and domain
reduction methods: The direct method case"
Anne Greenbaum, Courant Inst., "Optimal preconditioners of a given
sparsity pattern"
Willy Miranker, IBM - Yorktown Heights, "Memory intensive computation"
Hungwen Li, IBM - Yorktown Heights, "Sparse matrix vector
multiplication on polymorphic-torus"
--- A T T E N D E E S ---
Gerasoulis Apostolos Rutgers [email protected]
Greg Astfalk Convex Computer Corp. ihnp4!convex!astfalk
Josh Barnes Inst. Advanced Study [email protected]
Doug Baxter Yale [email protected]
Messaoud Benantar RPI [email protected]
Christian Bischof Cornell [email protected]
Rupak Biswas RPI [email protected]
Zaphiris Christidis IBM - Yorktown zaphiri@yktumh
Enrico Clementi IBM - Kingston -
Craig Douglas IBM - Yorktown [email protected]
Karen Dragon RPI [email protected]
Joe Flaherty RPI [email protected]
Stefano Foresti IBM - Kingston [email protected]
Robert van de Geijn Univ of Texas [email protected]
Bhaskar Ghosh Rutgers [email protected]
Gene Golub Stanford [email protected]
Anne Greenbaum Courant Inst [email protected]
Steve Hammond RPI [email protected]
Lars Hernquist Inst. Advanced Study [email protected]
Bo Kagstrom Univ of Umea [email protected]
Jacob Katzenelson MIT [email protected]
David Keyes Yale [email protected]
Shahin Khan FPS [email protected]
Kincho Lain RPI -
Hungwen Li IBM - Yorktown [email protected]
Ray Ludwig RPI [email protected]
Anita Mayo IBM [email protected]
Willard Miranker IBM - Yorktown wizard@yktvmv
Ravi Mirchandaney Yale [email protected]
Ramesh Natarajan IBM - Yorktown [email protected]
Isreal Nelken Rutgers [email protected]
David Pruett IBM - Kingston -
Laisal Saied Yale [email protected]
Joel Saltz Yale [email protected]
Dave Schneider IBM - Kingston [email protected]
Martin Schultz Yale [email protected]
Gautam Shroff RPI [email protected]
Mark Smith IBM [email protected]
Vijay Sonnad IBM - Lingston [email protected]
Wei-Pai Tang Univ of Waterloo [email protected]
Jordan Taylor IBM - Kingston -
Terbjorn Wiberg Univ of Umea [email protected]
Shahrokh Zargham Cornell [email protected]
Feng Zhao MIT [email protected]
------------------------------
From: Axel Ruhe <ruhe%cs.chalmers.sunet%[email protected]>
Date: 19 May 88 11:28 +0200
Subject: Professorship in Lund
Announcing a tenured chair as professor in
NUMERICAL MATHEMATICS
at the University of Lund Sweden.
Lund University founded 1668 has a rich academic profile.
Let me only mention that the successful applicant will get
distinguished colleagues in neighbouring subjects :
Lars Hormander, Mathematics
Georg Lindgren, Mathematical Statistics
Karl Johan Astrom, Automatic Control
Bjorn Roos, Quantum Chemistry
to mention a few.
Applications will be screened by 3 experts and decided by
the Swedish Government (Yes we are pompous here) on suggestion
of a committe from the faculty.
Deadline for application is June 1, 1988. A written application
addressed to the Government must have reached the University of
Lund at that time.
Postal address: Office of the registrar, Lund University, Box 117,
S-22100 Lund, Sweden
Street address: Paradisgatan 5, Lund
Telex: 33533 LUNIVER S
Telefax: int-46-46104720
The application should be supported by a c.v., list of publications,
copies of relevant publications and a short description of the
applicants research achievments and future plans. This supporting
material may arrive 3 weeks after the application date, and should
be sent in 4 copies.
Information can be obtained from Sten Henriksson
email: [email protected]
Telephone int-46-46108035
and of course from myself who graduated from Lund back in 1970.
Axel Ruhe
email: [email protected]
Telephone int-46-31721096 (office)
int-46-31483168 (home)
------------------------------
From: Tilak Ratnanather <NATILAK%[email protected]>
Date: 19-MAY-1988 17:25:04 GMT
Subject: Need Sparse SVD Code
I would like to know if there is a code to determine the SVD of a large sparse
matrix of order 729 with 85000 non-zero entries. This is an unsymmetric
matrix stored as a vector of values, column pointer and row start pointer.
Replies to me on [email protected] or Dan Howard on [email protected].
Use of the code will be gratefully acknowledged in the usual form.
Thanks
Tilak Ratnanather
Numerical Analysis Group
Oxford University Computing Lab
8-11 Keble Rd
Oxford
OX1 3QD
------------------------------
From: Hans Schneider <[email protected]>
Date: Fri, 20 May 88 15:48:26 cdt
Subject: LAA Special Issue on Matrix Valued Functions
LINEAR ALGEBRA AND ITS APPLICATIONS
Special Issue on
MATRIX VALUED FUNCTIONS
Recently there has been considerable progress in the subject
of matrix valued functions. This advance has been motivated
by the intrinsic mathematical interest in such functions as
well as their use in such applied areas as linear systems
theory and signal processing. The high level of current
activities in this subject is evident from the recent special
issues of LAA on Linear Systems and Control (volume 50 and a
forthcoming issue), the book Linear Algebra and its Role in
Systems Theory(AMS Contemporary Mathematics series, volume
47), the volume on Matrix Pencils (Springer Lecture Notes in
Mathematics, volume 973), a forthcoming issue of Operator
Theory : Advances and Applications, and from the perusal of
many mathematical and engineering journals.
Contributions for this special issue are invited on any
topic, theoretical or applied, in which matrix valued
functions and their structure play a significant role.
Besides linear systems theory and signal processing mentioned
above, these topics include numerical analysis(in particular
computation of structural invariants of matrix valued
functions), function and operator theoretic aspects of matrix
valued functions, and special classes of matrix valued
functions. Papers on operator valued functions will be
considered provided they have a finite dimensional flavor.
Papers should meet the usual publication standards of LAA and
will be refereed in the usual way. Both original research
papers and survey articles are invited. The deadline for
submission is July 1989 with expected publication in the
summer or fall of 1990. Papers may be sent to any of the
special editors of the issue:
Professor Joseph A. Ball Professor Leiba Rodman
Department of Mathematics Department of Mathematics
Virginia Polytechnic Institute College of William and Mary
Blacksburg, VA 24061 USA Williamsburg, VA 23185 USA
Dr. Paul Van Dooren
Philips Research Laboratory
2 Ave. van Becelaere
B-1170 Brussels, Belgium
------------------------------
From: Daniel Boley <[email protected]>
Date: Tue, 17 May 88 17:34:11 CDT
Subject: October Workshop on Iterative Methods
INVITATION TO ATTEND AND SUBMIT A PAPER TO
Workshop on Practical Iterative Methods for Large Scale Computations
Sun. Oct. 23 (evening) through Tues. Oct. 25, 1988
University of Minnesota, Minneapolis
Sponsored by the Minnesota Supercomputer Institute
The purpose of the workshop is to bring together computer scientists,
applied mathematicians, physical scientists, and engineers to present
their latest results and discuss their computational approaches for
solving the largest and most computational intensive problems with
iterative methods. The emphasis will be on large linear systems and
eigenvalue problems. Some of the applications that will be represented
are chemical reactions, device simulation, electron scattering, fluid
dynamics, reservoir simulation, and structural engineering; and the
algorithms of interest include conjugate gradient and Lanczos methods,
domain decomposition, multigrid techniques, and preconditioning.
ORGANIZING COMMITTEE
Daniel L. Boley, Youcef Saad, Donald G. Truhlar, and Robert E. Wyatt.
Current list of Speakers - Titles tentative
Daniel L. Boley, Minnesota, Parallel Algorithm for Eigenvalue Problems
Tony Chan, UCLA, Fourier Analysis of Preconditioners for the
Preconditioned Conjugate Gradient Method
William Coughran, AT&T Bell Labs, Iterative Methods in Semiconductor Simulation
Ernest R. Davidson, Indiana, Iterative Methods for Randomly Sparse Matrices
Howard C. Elman, Maryland, The Effect of Ordering on Parallel Iterative Solvers
Michel Fortin, Cit Universitaire, Quebec, Iterative Methods for Stokes and
Navier-Stokes Problems
Roland Glowinski, Houston, On the Iterative Aspects of Some Boundary Control
Problems
Gene H. Golub, Stanford, Banquet Speaker.
Anne Greenbaum, Courant Institute, Comparison of Linear System Solvers
Applied to Diffusion-Type Finite Element Equations
Roger Haydock, Oregon, Applications of the Recursion Method in Solid State
Physics
David Keyes, Yale, Domain Decomposition Methods for the Parallel Computation
of Reacting Flows
Donald J. Kouri, Houston, Robert E. Wyatt, Chemistry, Texas-Austin, and
Donald G. Truhlar, Minnesota, Iterative Methods for Large-Scale Quantum
Mechanical Scattering Calculations on Chemical Reactions
John G. Lewis, Boeing Computer Services, Recent Research in Iterative Methods
at Boeing
Mitchell Luskin, Minnesota, Relaxation and Gradient Methods for Molecular
Orientation in Liquid Crystals
Thomas Manteuffel, Los Alamos, The Role of Boundary Conditions ini
Preconditioning Elliptic Problems
Gerard Meurant, Centre d'Etudes de Limeil, Villeneuve St Georges, France,
Preconditioners for the Conjugate Gradient Method on Supercomputers
Antonio Navarra, Council of National Research, Modena, Italy, Applications
of Krylov Subspace Methods in Geophysical Fluid Dynamics Problems
C. M. M. Nex, Cambridge, United Kingdom, The Block Lanczos Algorithm and the
Calculation of Matrix Resolvents
Bahram Nour-Omid, Lockheed, Palo Alto, Applications of Lanczos ...
Allen H. Olson, Scripps Inst. of Oceanography, A Chebyshev Method for Solving
Huge Least Squares Problems
Gregory Parker, Oklahoma, and Joel D. Kress and Russell T Pack, Los Alamos,
Comparison of Lanczos and Subspace Iterations for Reaction Path Calculations
Beresford N. Parlett, U C Berkeley, Toward a Block Lanczos Code
Youcef Saad, Illinois-Urbana, Numerical Methods for Large Nonsymmetric
Eigenvalue Problems
Barry I. Schneider and Lee A. Collins, Los Alamos, Variation/Iteration Method
for the Solution of Scattering Problems
David S. Scott, Intel, Implementing Lanczos-like Algorithms on Hypercube
Architectures
Tayfun Tezduyar, Minnesota, Adaptive Implicit-Explicit and Element-by-Element
Iterative Schemes for Large-Scale Finite Element Computations
David Young, Texas-Austin, Opening Address of the Workshop.
CONTRIBUTED PAPERS & PROCEEDINGS
We welcome contributions to our featured poster session and/or to the
proceedings. An extended period Monday will be devoted to this featured
poster session. The proceedings will be published in a special issue
of Computer Physics Communications, a North-Holland journal.
All papers for the proceedings will be refereed by other participants.
DEADLINES
Poster abstracts (on special forms): July 22, 1988
Registration ($125 including preprints and many meals): July 22, 1988
Proceedings papers: August 22, 1988
REGISTRATION, ABSTRACT FORMS, DETAILED INFORMATION, and ALL SUBMISSIONS
Ms. Angie Vail
Workshop Administrator
Minnesota Supercomputer Institute
1200 Washington Ave. S.
Minneapolis, MN 55415
Phone: 612-624-1356
Lodging accommodations should be arranged directly with the hotel:
(We have a special rate for this workshop)
Whitney Hotel
150 Portland Ave.
Minneapolis, MN 55401
Phone: 612-339-9300.
------------------------------
From: Bernard Philippe <[email protected]>
Date: Wed, 18 May 88 08:51:55 +0200
Subject: Supercomputing Conference in France
Preliminary Program of the
1988 INTERNATIONAL CONFERENCE ON SUPERCOMPUTING
July 4-8, 1988
Saint-Malo, FRANCE
Sponsored by : ACM (SIGARCH)
In cooperation with : INRIA, IRISA, CSRD and IPSJ.
To receive the registration form, please contact :
Jocelyne ERHEL - IRISA
Campus de Beaulieu
35042 RENNES Cedex
FRANCE
Telex : UNIRISA 950 473F - Fax : (33) 99 38 38 32
e-mail : [email protected]
MONDAY, 4th JULY 1988
MORNING
OPENING SESSION
KEYNOTE ADDRESS : Why Supercomputers ?
Jacques-Louis Lions, Professor, College de France
and President, Centre National d'Etudes Spatiales
SESSION 1A : OPERATING SYSTEMS ISSUES
. Page Table Management in Local/Remote Architec-
tures
Mark A. Holliday, Duke University, USA
. Measurement and Prediction of Contention in
Multiprocessor Operating Systems with Scientific
Application Workloads
George E. Bier and Mary K. Vernon, University of
Wisconsin, USA
SESSION 1B : OPTICAL INTERCONNECTIONS
. OPTIMUL : An Optical Interconnect for Multipro-
cessor Systems
Norman S. Matloff, Stephen Kowel, and Charles
Eldering, University of California at Davis, USA
. Optical Link in the Delft Parallel Processor
L. Dekker, E.E.E. Frietman, W. Smit and J.C.
Zuidervaart, University of Technology, Delft,
The Netherlands
MONDAY, 4th JULY 1988
AFTERNOON
INVITED PRESENTATION : A Robust Parallel Solver
for Sparse Linear Systems
Ahmed Sameh, Center for Supercomputing Research &
Development, University of Illinois, USA
SESSION 2A : SPECIAL PURPOSE COMPUTERS
. PERCOLA : A Special-Purpose Programmable 64-bit
Floating-Point Processor
J.M. Normand,Centre d'Etudes Nucleaires, Saclay,
France
. Multiple-transform Pipelines for Image Coding
Anna Antola, Politecnico di Milano, Italy
. System Architecture of Parallel Processing
System -Harray-
Hayato Yamana, Toshikazu Marushima, Takashi
Hagiwara and Yoichi Muraoka, Waseda University,
Japan
. DYPP : A VLSI Dynamic-Graph Ensemble Machine
Marius V.A.Hancu,Bell Northern Research,Canada ;
Kenneth C. Smith, University of Toronto, Canada
. BVE : A Wafer-Scale Engine for Differential
Equation Computation
Jose G. Delgado-Frias,University of Oxford, UK ;
Douglas M. Green, Dowell-Schlumberger, UK
SESSION 2B : ENVIRONMENTS
. From Mathematical Specifications to Parallel
Programs on a Message-Based System
Th. Ruppelt and G. Wirtz, Rheinische Friedrich-
Wilhelm Universitt, FRG
. Interactive Program Improvement Via EAVE : an
Expert Adviser for Vectorization
Pradip Bose, IBM Thomas J. Watson R.C., USA
. Parasight : A High-Level Debugger/Profiler
Architecture for Shared-Memory Multiprocessors
Ziya Aral and Ilya Gertner, Encore Computer, USA
. The Automated Crystal Runtime System : A
Framework
Joel H. Saltz, Ravi Mirchandaney, Roger M. Smith
and Kay Crowley, Yale University, USA ;
David M. Nicol, College of William and Mary, USA
TUESDAY, 5th JULY 1988
MORNING
INVITED PRESENTATION : Who Needs Supercomputers ?
Mike Farmwald, Center for Supercomputing Research
& Development, University of Illinois, USA
SESSION 3A : COMMERCIAL SYSTEMS
. The "Bussing Freeway": A Technological
Breakthrough
Hanan Potash and Steve L. Adams, Scientific
Computer Systems, USA
. The Gould NP1 System Interconnection
Duc J. Vianney, James H. Thomas and Vicki
Rabaza, Gould, USA
. Cluster Oriented Architecture for the Mapping of
Parallel Processor Networks to High Performance
Applications
Falk D. Kbler, Parsytec, FRG
. Micro-Analysis of the Titan's Operation Pipe
John Sanguinetti,Ardent Computer Corporation,USA
. High-speed Processing Schemes for Summation Type
and Iteration Type Vector Instructions on
HITACHI Supercomputer S-820 System
Hideo Wada, Koichi Ishii, Masakazu Fukagawa and
Shun Kawabe, Hitachi, Japan
SESSION 3B : DETECTING PARALLELISM
. Determining Useful Parallelism
Fran Allen, Michael Burke, Ron Cytron, Jeanne
Ferrante and Vivek Sarkar, IBM Thomas J. Watson
R.C., USA
Wilson Hsieh, Massachusetts Institute of
Technology, USA
. Partitioning Programs for Parallel Execution
Milind Girkar, University of Illinois, USA
. Requirements for Optimal Execution of Loops with
Tests
Augustus K. Uht, University of California at San
Diego, USA
. On the Problem of Optimizing Parallel Programs
for Hierarchical Memory Systems
Dennis Gannon, Indiana University, USA
TUESDAY, 5th JULY 1988
AFTERNOON
INVITED PRESENTATION : Supercomputing in Control
and Fluid Dynamics
Roland Glowinski, University of Houston, USA, and
INRIA, France
SESSION 4A : SUPERCOMPUTER APPLICATIONS
. Parallelization and Performance Evaluation of
Circuit Simulation on a Shared Memory Multipro-
cessor
P. Sadayappan, Ohio State University, USA
V. Visvanathan, AT&T Bell Laboratories, USA
. Three-Dimensional Numerical Simulations of the
Czochralski Bulk Flow on a CRAY X-MP
Multiprocessor Architecture
Wolfgang E. Nagel and Kurt Wingerath,
Kernforschungsanlage Jlich, FRG
. Tests on Parallel Machines of a Domain Decomposi-
tion Method for a Composite Three-Dimensional
Structural Analysis Problem
Franois-Xavier Roux, O.N.E.R.A., France
. Block-Iterative Finite Element Computations for
Incompressible Flow Problems
T.E. Tezduyar and J. Liou, University of
Minnesota, USA
R. Glowinski,T. Nguyen and S. Poole, Houston,USA
. Performance Evaluation for an Ocean General
Circulation Model:Vectorization and Multitasking
Patrick Andrich and Gurvan Madec, Universite de
Paris VI, France
SESSION 4B : MEMORY SYSTEMS
. Performance Study of a Clustered Shared-Memory
Multiprocessor
Keki B. Irani and Ahmed R. Naji, University of
Michigan, USA
. Performance of a Shared Memory System for
Vector Multiprocessors
Stephen W. Turner and Alexander V. Veidenbaum,
University of Illinois, USA
. A Two-Tier Memory Architecture for High-
Performance Multiprocessor Systems
Tan M. Nguyen,Vason P. Srini and Alvin M.
Despain, University of California at Berkeley,
USA
. A Cache Coherence Approach For Large
Multiprocessor Systems
James K Archibald,Brigham Young University, USA
. Hierarchical Registers for Scientific Computers
John A. Swensen and Yale N. Patt, University of
California at Berkeley, USA
WEDNESDAY, 6th JULY 1988
MORNING
INVITED PRESENTATION : Why is the Supercomputer
Debate on Technology when the Future is with
Software ?
Carl Ledbetter, President, ETA Systems, USA
SESSION 5A : HYPERCUBES
. Hyperswitch Network for the Hypercube Computer
E. Chow, H. Madan and J. Peterson, Jet
Propulsion Laboratory, USA
D. Grunwald and D. Reed, University of Illinois,
USA
. Performance Evaluation of a Formally
Supercomputer-based Monte Carlo Program on a
T800 Transputer Network
M. Luckas, Universitat Duisburg, FRG
. A Parallel Spline Collocation-Capacitance Method
for Elliptic Partial Differential Equations
C.C. Christara, E.N. Houstis and J.R. Rice,
Purdue University, USA
. Pipelined Data Parallel Algorithms - Concept and
Design
Chung-Ta King, Wen-Hwa Chou and Lionel M. Ni,
Michigan State University, USA
SESSION 5B : ADVANCED COMPILING TECHNIQUES
. Introducing Symbolic Problem Solving Techniques
in the Dependence Testing phases of a Vectorizer
A. Lichnewsky and F. Thomasset, INRIA, France
. The Importance of Direct Dependences for
Automatic Parallelization
Thomas Brandes, Universitat Marburg, FRG
. Dependence of Multi-Dimensional Array References
D.R. Wallace, Alliant Computer Systems, USA
. Array Expansion
Paul Feautrier, Universite Paris VI, France
WEDNESDAY, 6th JULY 1988
AFTERNOON
Visit of Le Mont Saint-Michel and its abbey.
THURSDAY, 7th JULY 1988
MORNING
SESSION 6A : CODE GENERATION AND OPTIMIZATION
. Code Scheduling and Register Allocation in Large
Basic Blocks
James R. Goodman and Wei-Chung Hsu, University
of Wisconsin, USA
. Optimization of Horizontal Microcode Generation
for Loop Structures
Christine Eisenbeis, INRIA, France
. Optimizing Horizontal Microprograms for
Vectorial Loops with Timed Petri Nets
C. Hanen, Universite de Paris VI, France
SESSION 6B : PARALLEL ALGORITHMS
. Parallel and Vector Conjugate Gradient-like
Algorithms for Sparse Nonsymmetric Linear
Systems
C. Radicati di Brozolo and Y. Robert, IBM ECSEC,
Italy
. Arnoldi-Tchebychev for Large Scale Matrices and
Its Vectorizability
Diem Ilo, F. Chatelin and J.C. Patau, IBM,France
. Boundary Integral Domain Decomposition on
Hierarchical Memory Multiprocessors
E. Gallopoulos and D. Lee, University of
Illinois, USA
SESSION 7 : PERFORMANCE MEASUREMENTS
. A Benchmark Package for Sparse Matrix
Computations
Youcef Saad and Harry A.G. Wijshoff, University
of Illinois, USA
. An Evaluation of Cray-1 and Cray X-MP
Performance on Vectorizable Livermore Fortran
Kernels
Ju-ho Tang and Edward S. Davidson, University of
Michigan, USA
. Performance Evaluation of Static and Dynamic
Memory Systems on the Cray-2
D.A. Calahan, University of Michigan, USA
THURSDAY, 7th JULY 1988
AFTERNOON
INVITED PRESENTATION : Future Directions of High-
Performance Scientific and Engineering Computing
Gordon Bell, ARDENT Computer Corporation, USA
INVITED PRESENTATION : Yes, We can map
Computations onto a Parallel Computer Efficiently
H.T. Kung, Carnegie-Mellon University, USA
SESSION 8 : PANEL : Recent Advances and Futures
Trends
Session chairman : David J. Kuck, CSRD and
University of Illinois, USA
Cocktail in the Casino of Dinard, on the other
bank of the Rance estuary (crossing in motorboat)
FRIDAY, 8th JULY 1988
MORNING
INVITED PRESENTATION:Parallel Applications on RP3
George Paul, IBM, Thomas J. Watson R.C., USA
SESSION 9A : SYMBOLIC PROGRAMMING
. PARCEL : Project for the Automatic Restructuring
and Concurrent Evaluation of Lisp
W. Ludwell Harrison,III and David A. Padua,
University of Illinois, USA
. Vectorization Techniques for Prolog
Yasusi Kanada, Keiji Kojima and Masahiro Sugaya,
HITACHI, Japan
. A Hybrid Scheme for Detecting AND-Parallelism in
Prolog Programs
Hong Xia and Wolfgang K. Giloi, GMD-FIRST and TU
Berlin, FRG
. Lazy Evaluation and the Logic Variable
Keshav K. Pingali, Cornell University, USA
SESSION 9B : SYNCHRONIZATION I
. An Approach to Synchronization for Parallel
Computing
V.P. Krothapalli and P. Sadayappan, Ohio State
University, USA
. Generating Sequential Code from Parallel Code
J. Ferrante, M. Mace and B. Simons, IBM, USA
. Impact of Self-Scheduling Order on Performance
of Multiprocessor Systems
P. Tang, P.C. Yew and C.Q. Zhu, University of
Illinois, USA
FRIDAY, 8th JULY 1988
AFTERNOON
INVITED PRESENTATION : Using Silicon and Gallium
Arsenide Technologies for New Supercomputer Design
Stephen Nelson, CRAY Research, USA
SESSION 10A : SYNCHRONIZATION II
. Towards a Large Number of Pipeline Processors in
a Tightly Coupled Multiprocessor Using no Cache
A. Seznec and Y. Jegou, IRISA/INRIA, France
. A Simple Solution to Lamport's Concurrent
Programming Problem with Linear Wait
B.K. Szymanski,Rensselaer Polytechnic Institute,
USA
. Synchronization Using Counting Semaphores
Vivek Sarkar, IBM Thomas J. Watson, USA
SESSION 10B : PROCESSOR ARRAYS
. Givens Elimination on Systolic Arrays
B. Louka and M. Tchuente, Faculte des Sciences
de Yaounde, Cameroon
. Nonlinear Network Optimization on a Massively
Parallel Connection Machine
S. A. Zenios, University of Pennsylvania, USA
R. A. Lasken, Perkin-Elmer, USA
. Ray Tracing on a Connection Machine
H. C. Delany, MIT Media Laboratory, USA
------------------------------
End of NA Digest
**************************
-------