gannon@iuvax.cs.indiana.edu (Dennis Gannon) (01/03/91)
Third ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming.
Williamsburg, Virginia, April 21-24, 1991
Preliminary Program, Tutorial Announcement and Registration Forms
PPoPP
General Chair: Dennis Gannon, Indiana University
Program Chair: Jeanne Ferrante, IBM Research
Program Committee:
David Padua, Illinois David Callahan, Tera Computer
Elizabeth Williams, SRC Jack Dongarra, Tennessee
Marina Chen, Yale Piyush Mehrotra, ICASE
Larry Snyder, Washington Donna Bergmark, Cornell NSF
Anthony Hey, Southhampton Francine Berman, UCSD
James Larus, Wisconsin Keshav Pingali, Cornell
Local Arrangements Chair: Piyush Mehrotra,
ICASE NASA Langley Research Center Hampton, Va.
(804) 864-6134
________________________________________________________________
Preliminary Program
_________________________________________________________________
Monday, April 22
Introductory Remarks: Robert Voigt, Director, ICASE NASA
Langley Research Center.
Keynote Address: Marc Snir, IBM Research.
"The right programming languages for parallel machines -- does
complexity theory offer any useful insight?"
Monday Session 1: Chair: David Callahan
Optimal Schedules for Parallel Prefix Computation with Bounded
Resources
Alexandru Nicolau, University of California, Irvine,
Haigeng Wang, University of California, Irvine
Parallel-Program Transformation Using a Metalanguage
J. Allan Yang, Young-il Choo, Yale University
Mapping Concurrent Programs to VLIW Processors
Hester Bakewell, Donna J. Quammen,
P. Y. Wang, George Mason University
Monday Session 2: Chair: Keshav Pingali
A Unified Framework for Systematic Loop Transformations
Lee-Chung Lu, Yale University
Scanning Polyhedra with DO Loops
Corinne Ancourt, Francois Irigoin, Ecole Nationale
Superieure des Mines de Paris
Removal of Redundance Dependences in DOACROSS Loops with
Constant Dependences
V. P. Krothapalli, University of Wisconsin Oshkosh,
P. Sadayappan, The Ohio State University
Monday Session 3: Chair: Anthony J.G. Hey
Exploitation of APL Data Parallelism on a Shared Memory MIMD
Machine
Dz-ching Ju, University of Texas at Austin,
Wai-Mee Ching, IBM T.J. Watson Research Center
A Production-Quality C* Compiler for a Hypercube Multicomputer
Philip J. Hatcher, Anthony J. Lapadula, Robert R. Jones,
University of New Hampshire,
Michael J. Quinn, Ray J. Anderson, Oregon State University
April 23
Tuesday Session 4: Chair: James R. Larus
Andorra-I: A Parallel Prolog System that Transparently Exploits both
And- and Or-Parallelism
Vitor Santos Costa, David H. D. Warren, Rong Yang,
University of Bristol
Coarse-Grain Parallel Programming in Jade
Monica S. Lam, Martin Rinard, Stanford University
Tuesday Session 5: Chair: Elizabeth Williams
Scalable Reader-Writer Synchronization for Shared-Memory
Multiprocessors
John M. Mellor-Crummey, Rice University
Michael L. Scott, University of Rochester
Dynamic Node Reconfiguration in a Parallel-Distributed Environment
Michael J. Feeley, Brian N. Bershad, Jeffrey S. Chase,
Henry M. Levy, University of Washington
Exploiting Operating System Support for Dynamic Page Placement on
a NUMA Shared Memory Multiprocessor
Richard P. LaRowe Jr., James T. Wilkes, Carla Schlatter Ellis,
Duke University
Tuesday Session 6: Chair: Donna Bergmark
Improving the Accuracy of Data Race Detection
Robert H. B. Netzer, Barton P. Miller,
University of Wisconsin-Madison
Race Frontier: Reproducing Data Races in Parallel-Program Debugging
Jong-Deok Choi, Sang-Lyul Min, IBM Research.
Optimistic Parallelization of Communicating Sequential Processes
David F. Bacon, Robert E. Strom, IBM Research
Panel: Problems and Issues in Parallel C Programming.
Chair: Dennis Gannon
April 24
Wednesday Session 7: Chair: Marina Chen
Parallelizing a New Class of Large Applications over High-speed
Networks
H. T. Kung, Peter Steenkiste, Marco Gubitoso,
Manpreet Khaira, Carnegie Mellon University
Parallelization and Performance of Conjugate Gradient Agorithms on
the Cedar Hierarchical-Memory Multiprocessor
Ulrike Meier, Rudolph Eigenmann, University of Illinois
Wednesday Session 8: Chair: Fran Berman
The Integration of Application and System Based Metrics in a Parallel
Program Performance Tool
Jeffrey K. Hollingsworth, R. Bruce Irvin, Barton P. Miller,
University of Wisconsin-Madison
Event-Based Performance Perturbation
Allen D. Malony, CSRD, University of Illinois
A Static Performance Estimator to Guide Data Partitioning Decisions
Vasanth Balasundaram, IBM T.J. Watson Research Center,
Geoffrey Fox, Syracuse University,
Ken Kennedy, Ulrich Kremer, Rice University
____________________________________________________________________
TUTORIALS
___________________________________________________________________
Sunday April 21
A: Compiling for Parallelism
8:00 am to 12:00, Ron Cytron, IBM Research.
This tutorial will examine the automatic restructuring of programs for
architectures that support various forms of concurrent execution. The
topics considered include: data and control dependence analysis
techniques for discovering parallelism in sequential programs,
restructuring techniques for improving the effectiveness of such analysis,
and architectural-specific transformations for exploiting diverse
concurrency features such as vectors, multiprocessors, and wide
instruction words. The tutorial will include recent work on management
of memory hierarchies and analysis techniques for shared-memory
parallel programs.
B: Data parallel programming: Programming Primitives and
Performance.
1:00 pm. to 5:00 pm. S. Lennart Johnsson Harvard University.
Data parallel programming implies programming in a language with an
array syntax, such as in the proposed Fortran 9X, or CM-Fortran, C*, or
*Lisp. We will review these languages, and focus on examples that
illustrate useful idioms for programming data parallel architectures.
Massively parallel architectures are driving the development of languages
with an array syntax, while performance is a driving force for massively
parallel architectures. We will emphasize performance issues in data
parallel programming and give examples of the importance of the proper
choice of algorithms, data allocation, and data motion for good
performance. The benefit of a small set of communications functions for
portability of programs with respect to performance will be illustrated.
C: Building Parallel Programs
1:00 pm. to 5:00 pm. David Gelernter, Yale University.
Parallel programming environments that are high-level and efficient and
portable and supported by decent tools are widely available and
commercially supported. Consequently, research has shifted upwards a
level: how do we use these tools methodically to develop clear and
efficient parallel programs?
We will present a method for parallel programming based on a three-
way categorization of the basic program structures for parallelism. I'll
describe the logical basis and practical application of the method, give
some examples and discuss their performance. The method applies in any
asynchronous parallel environment (We will discuss performance on
shared- and distributed-memory multiprocessors and on conventional
LANs). We will use a combination of C and (the coordination language)
Linda in presenting the method and examples.
We will conclude by looking at some of the broader programming
considerations---including issues of modularity and clarity, management
of physically dispersed resources, heterogeneity and data persistence---
that have made ensemble programming in general the emerging
centerpoint of systems research.
D: Tools, Languages and Environments for Fragmented Memory
MIMD Multiprocessors
8:00 am.- 12:00 Joel Slatz, ICASE, NASA Langley.
This tutorial will provide a comprehensive survey of systems and methods
designed to ease the burden borne by those who program distributed
memory MIMD machines such as the iPSC, NCUBE and transputer based
architectures. Compilers for distributed memory architectures attempt to
provide the user with an illusion of shared memory. These compilers
orchestrate the distribution of data and work. We will describe how such
compilers function and how such compilers are able to handle a range of
different applications. The compiler efforts at Pacific Sierra, ICASE,
Rice, Yale, the University of Colorado, and University of Vienna will be
reviewed. Shared virtual memory methods use operating system methods
to support a shared name space. We will describe the shared virtual
memory efforts at Princeton, Carnegie Mellon University, Intel and
IRISA, outlining the potential advantages and drawbacks of such
methods.
We will also describe a collection of tools and environments such as
high level communication primitives, tools and environments used for
performance measurement and tuning, programming environments aimed
at specific classes of applications, and dense and sparse linear algebra
library procedures.
__________________________________________________________
CONFERENCE REGISTRATION
Name ____________________________________________________________
Prefered Name on Badge __________________________________________
Affiliation _____________________________________________________
Address _________________________________________________________
_________________________________________________________________
Phone: (_______)_________________________
FAX Number:(______)______________________
Email ___________________________________
Program Registration Fee:
Early (by 3/20) Late (after 3/20)
Non-Members $300 $350
ACM Member $275 $325
Full Time Student $150 $200
ACM number _________________________________
For students: school name and student identification number
________________________________________________________
Program registration includes: Sunday night reception, Monday and
Tuesday lunch, Monday night dinner, coffee breaks and proceedings
(except for student registration which does not include the reception and
dinner).
Additional dinner tickets @ $35 each: number _______
Tutorial(s) registration:
Circle one or two tutorials: A(am) B(pm) C(pm) D (am)
Early (by 3/20) Late (after 3/20)
Non-ACM ACM Student Non-ACM ACM Student
Any one: $150 $125 $100 $175 $150 $125
Any two: $250 $200 $150 $300 $250 $200
Tutorial Registration includes one copy of the notes and, for two tutorials,
it includes lunch.
Please pay by check or money order payable to:
ACM/SIGPLAN PPoPP 91.
Please mail conference registration form and check to:
PPoPP 91 c/o Emily Todd
ICASE MS 132C,
NASA Langley Research Center
Hampton, VA 23665
Phone: 804 864 2175 FAX: 804 864 6134
email emily@icase.edu
Refund requests must be made in writing and postmarked before March
20, 1991. Registrations can be transferred by giving a Substitute for the
conference registration and confirmation receipt.
____________________________________________________________________
PPoPP 91 HOTEL REGISTRATION
Please mail to: Williamsburg Hilton, 50 Kingsmill Road, VA 23185 USA
Call Reservations (804)-220-2500 . (For telephone registration please
mention group name PPoPP91.) Deadline March 20, 1991. Fax: (804)-
220-2500 ext 7601
Group Name: PPoPP91
Name ______________________________________________________________
Affiliation ________________________________________________________
Address ____________________________________________________________
____________________________________________________________________
Daytime Phone Number (______)_______________________________________
Please include country, area, city code where appropriate
Arrival Date __________________ Arrival time _____________
Departure Date ________________ Departure time ___________
Sharing with _____________________________________________
Please circle desired room:
Single: $68 (also with spouse)
Double: $74 (with another conference attendee)
Note that these rates do not include VA tax of 6.5%. These rates are also
offered to attendees arriving two days prior to start of the conference and
for two days after the conference.
Check-in time is 3:00 pm; check-out time is 12:00 noon. Arrivals after
6:00 pm must guarantee first night accommodation with check, money
order, or major credit card.
Credit card company ________________________________________________
Card number ________________________________________________________
Expiration date ____________________________________________________
Signature __________________________________________________________
Reservations after the contracted block of rooms is full or cut off date of
March 20 are subject to space and rate availability.
TRANSPORTATION: Williamsburg, Virginia is 15 miles from Newport
News/Williamsburg International Airport in Newport News, VA and 45
miles from Byrd International Airport in Richmond and Norfolk
International Airport. For ground transportation from the airport to the
hotel the following limousine services are available:
Byrd Airport (Richmond) - Groome Transportation (804) 222-7222.
Leaves every hour on the hour. $20/one person; $13/two or more.
Norfolk Airport - Airport Limousine Service (804) 857-1231.
Leaves every hour on the half hour. $21/one person; $14/two or more.
Newport News - Williamsburg Airport - Williamsburg Limousine (804) 877-0297
Reservations must be made in advance. $15 per person.
INFO ABOUT WILLIAMSBURG:
Virginia's Historic Triangle of Williamsburg, Jamestown and Yorktown is a
delicate blend of the past and the present. The restoration of Colonial
Williamsburg has set a standard of excellence unmatched anywhere in the
world. Its gardens, architecture, furnishings and fine arts collections
reflect life in Williamsburg when this small city was the social, cultural
and political capital of England's largest colony in the New World.
The scenic Colonial Parkway, winding through over 9,000 acres between the
James and York Rivers, connects Williamsburg with Jamestown and Yorktown.
Eight miles west of Williamsburg, on Jamestown Island, where America's
first permanent English settlement was established in 1607. Twenty-three
miles east of Williamsburg are the now silent battlefields of Yorktown.
Along the James River, the magnificant grounds and exquisite antiques of
historic plantations still reflect early southern plantation life.
The weather in Williamsburg ranges from 65 - 75 degrees in April.