home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!spool.mu.edu!olivea!tekbspa!denny
- From: nikola@prosun.first.gmd.de (Nikola Serbedzija)
- Newsgroups: news.announce.conferences
- Subject: Massively Parallel Programming Models (MPPM)
- Message-ID: <Bzqo4G.A16@tss.com>
- Date: 24 Dec 92 01:00:15 GMT
- Expires: Mon, 1 Feb 1993 08:00:00 GMT
- Sender: denny@tss.com (Denny Page)
- Organization: GMD-FIRST, D-1000 Berlin 10
- Lines: 132
- Approved: denny@tss.com
-
-
- CALL FOR PAPERS AND/OR PARTICIPATION
-
-
- Working Conference on
-
- MASSIVELY PARALLEL PROGRAMMING MODELS:
-
- Suitability, Realization, and Performance
-
-
- September 20-23, 1993
- Glienecke Hunting Palace, Berlin, Germany
-
- Organized in Cooperation with IEEE Computer Society
- and Gesellschaft fuer Informatik
-
-
- The next generation of supercomputers will be massively parallel,
- distributed memory MIMD architectures with thousands or ten thousands
- of nodes. User acceptance demands that such systems should be
- programmable in an appropriate, easy-to-use, application-oriented style
- which, nevertheless, ensures the highest possible program efficiency.
- The currently dominating message-passing programming paradigm hardly
- meets those requirements. One approach to avoid explicit message passing
- in the program is the use of run time libraries that carry out the
- communication. Another, architectural solution is the virtual shared
- memory that provides the user with a shared memory programming paradigm,
- hiding the actual message passing in the bowels of the system. The
- current problem with virtual shared memory is unsatisfactory performance.
- A better solution could be to have a parallelizing compiler that is
- smart enough to take a sequential program, partition it into
- communicating processes with the goal of optimal workload distribution,
- and inserts the required inter-process communication constructs. At this
- point in time it is very much debated whether such a compiler will ever
- be feasible. Last but not least, novel, more appropriate programming
- models for massively parallel systems may evolve that will avoid the
- difficulties with the existing paradigms.
-
- The purpose of the working conference is to bring together computer
- architects, software engineers, and supercomputer users with experience
- in highly parallel applications and discuss what is needed from the
- user point of view and how to meet the user demands. Another question
- is how efficiently can the envisioned solutions be implemented.
- Experiences with prototype implementations of virtual shared memory
- systems, parallelizing compilers, and existing as well as novel
- programming models should be presented, discussed, and evaluated.
- Therefore, equal time will be dedicated to the presentation of papers
- and discussions. This will hopefully lead to a consensus as to what
- the appropriate programming model(s) for massively parallel systems
- should be. The proceedings (papers and summaries of discussions) will
- be published as a book in the Springer LNCS Series.
-
- Participation in the working conference is by invitation. Please apply
- -- with or without an interesting paper.
-
- --------------------------------------------------------------------------
- Conference Data
-
-
- Short Title: Working Conference on Massively Parallel
- Programming Models (MPPM)
-
- Location: Berlin, Germany, Glienicke Hunting Palace
-
- Time: September 20-23, 1993
-
- Conference Fee: $ 650 for non-members of IEEE or GI
- $ 500 for members of IEEE-CS or GI
-
- How to apply: Please send an extended abstract to one of the
- program chairpersons or one of the program committee
- members no later than February 1, 1993.
-
- Notification date: You will be notified whether your papers has
- been accepted before April 1, 1993.
-
- Paper due: A preliminary version of the paper must be submitted
- no later than August 1, 1993. This version will be
- copied and distributed to the participants. The size
- of the paper is limited to 10 pages (single-spaced,
- in Time or similar font of 10 points or larger) The
- precise format instructions will be sent with the
- acceptance notification.
-
-
-
- Program Committee:
- Prof. Wolfgang K. Giloi, Co-Chairman
- Prof. Stefan Jahnichen, Co-Chairman
- Dr. Bruce D. Shriver, Co-Chairman
-
- Arvind, MIT, U.S.A.
- Guang Gao, McGill University, Canada
- Erol Gelenbe, Universite de Paris, France
- John Gurd, University of Manchester, UK
- Anthony Hey, University of Southampton, UK
- William Jalby, INRIA and University of Rennes, France
- Ted Lewis, Oregon State University
- Mario Tokoro, Sony, Japan
- Roland Vollmar, University of Karlsruhe, Germany
- Hans Zima, U. Wien, Austria
-
-
- Addresses:
- Professor Wolfgang K. Giloi
- GMD FIRST
- Rudower Chausee 5
- Building 13.7
- 1199 Berlin-Adlershof
- Germany
- Phone: 49-30-6704-2659
- Fax: 49-30-6704-5610
- e-mail: w.giloi@compmail.com
-
- Professor Stefan Jahnichen
- TU-Berlin
- Institut fur Angewandte
- Informatik, Softwaretechnik
- FR 5-6
- Franklinstrasse 28/29
- D-1000 Berlin 10
- Phone: 49-30-314732-31
- Fax: 49-30-314734-88
- e-mail: jaehn@cs.tu-berlin.de
-
- Dr. Bruce D. Shriver
- 17 Bethea Drive
- Ossining, NY 10562
- U.S.A.
- Phone and fax: 1-914-941-9181
- e-mail: b.shriver@compmail.com
-