Newsgroups: comp.parallel.mpi,comp.parallel.pvm
From: mpcs@emess.mscd.edu (MPCS)
Subject: MPCS '98 Keynote
Organization: Metropolitan State College of Denver	
Date: 10 Oct 1997 20:43:31 GMT
Message-ID: <61m41j$rjc$2@clem.mscd.edu>

                              MPCS 1998 Keynote

                    The Beginnings of Massive Parallelism

                              David H. Schaefer

Abstract 

The Massively Parallel Processor (the MPP),which was in operation at NASA's
Goddard Space Flight Center from 1983 to 1991, had much greater parallelism
than had ever before been attained, and introduced the term "massive
parallelism".  The computer contained 16,384 processing elements.  It is
now a part of the collection of the Smithsonian Institution.

The MPP emerged from experiments performed at Goddard in the 1960's that
probed classical optical methods of performing parallel computations.  From
these experiments developed the idea of building very high speed computers
with discrete electronic parallel logic elements that were connected to
each other through fiber optic bundles. These elements would performing
sixteen thousand operations simultaneously. The basic data entity of these
computers (as in any parallel computer) was an image, and as the basic data
entity in the Chinese language is an image, we named our envisioned
computers Tse computers after the Chinese character for a Chinese
character.  In the talk some of the parallel hardware developed will be
shown.

Other efforts to fabricate parallel computers were also underway in the
1960's.  In fact as early as 1958 S. H. Unger envisioned  "A stored program
computer which can handle spatial problems by operating directly on
information in spatial form without scanning or using other techniques for
scanning the problem into some other domain."  One of Unger's hopes was
that his computer could play chess. The architecture of this computer is
interesting as it had the ability to send data from processing element to
processing element in an unclocked manner.  This feature was later
incorporated into the CLIP IV computer in England.

There have been questions about Daniel Slotnick's SOLOMON computer.  It can
be documented that in 1963 a three by three prototype of the computer was
fabricated. There is no record of larger SOLOMON computers being produced.
A penciled note on a report about the prototype says  "1st PDE solution,
Feb.  18, 1963."   That date may well mark the running of the world's first
parallel program.
---------------------------------------------------------------------------
Professor David H. Schaefer has been an Associate Professor of Electrical
and Computer Engineering at George Mason University since 1981. Previous to
joining George Mason, he was Head of the Computer Development Section at
NASA's Goddard Space Flight Center.  There he managed a program to develop
advanced methods of processing data on board spacecraft, and he initiated
and managed the program to build the Massively Parallel Processor (MPP),
the first U.S. massively parallel computer.  The MPP is a SIMD computer
with 16,384 processing elements.  This computer is now a part of the
collection of the Smithsonian Institution.  He is the author or co-author
of over 30 journal and proceedings papers, has articles in seven books, and
holds twelve patents for computer-related inventions. His major research
interest is in the field of advanced non-traditional computer architectures.
Present research involves the investigation of Petri net modeling of
computers. 
---------------------------------------------------------------------------
MPCS'98 is the third bi-ennial conference on a wide variety of subjects
related to parallel processing.  See the web page at: http://www.mpcs.org
for more information and on how to submit a paper, which are due by
November 7th, 1997.