[comp.parallel] Thinking Machines Announcement

fpst@hubcap.clemson.edu (Steve Stevenson-Moderator) (11/30/89)

[ This was passed on to me by Bob Panoff. It was distributed at NRL.
	- Steve
]

   Washington, Nov. 28, 1989 -- Thinking Machines Corporation announced
   today the receipt of a $12 million contract from the Defense Advanced
   Research Projects Agency (DARPA) to accelerate its development of the
   world's most powerful supercomputer, with peak speeds above one
   trillion operations per second (1 TeraOps).  

   Under the terms of the contract, Thinking Machines will produce a
   supercomputer 1,000 times more powerful than today's supercomputers.
   "Our customers have told us that they need this level of performance,"
   explained Danny Hillis, co-founder and Chief Scientist of Thinking
   Machines.  "And we feel we're in the best position to provide it."

   A September 8, 1989 report from the Executive Office of Science and
   Technology Policy stated that, "High performance computing is a vital
   and strategic technology, exerting strong leverage on the rest of the
   computer industry and other cutting-edge areas....  We cannot afford
   to cede our historical leadership in high performance computing and
   its applications."  The report detailed target applications for this
   new level of supercomputing, including world climate prediction,
   semiconductor circuit design and test, superconductivity research, and
   mapping the human genome to support advanced drug design (see attached
   chart).

   The project, code-named MEGA, builds on the proven massively parallel
   architecture of the company's Connection Machine supercomputer,
   currently installed at more than 40 scientific, industrial and
   commercial sites throughout the world.  Installations include Los
   Alamos National Laboratory, the National Center for Atmospheric
   Research, NASA Ames Research Center, Dow Jones Information Services,
   United Technologies and Lockheed Aerospace.

   Chart inserted here.  Title: Some Grand Challenges and Their Projected
   Computational Requirements.  Source: Office of Science and Technology
   Policy.  Chart shows that applications such as Vision, Climate
   Modeling, Fluid Turbulence, Human Genome, Vehicle Dynamics, Ocean
   Circulation, Viscous Fluid Dynamics, Semiconductor Modeling,
   Superconductor Modeling, and Quantum Chromodynamics require 10 Gwords
   of Memory Size and 1 Tflops of Processing Speed.

   Thinking Machines Corporation was formed to apply parallel processing
   techniques to the growing number of data-intensive computing
   applications in business and science.  Since announcing the Connection
   Machine system in 1986, the company has grown to become the second
   largest supercomputer supplier in the U.S., with 10 percent of the
   world's installed base.  Thinking Machines is headquartered at 245
   First Street, Cambridge, Mass.

   For further information, contact:

   Alison Holland, Thinking Machines Corporation, (617) 876-1111 or Maura
   FitzGerald/Susan Curtin, Miller Communications, (617) 536-0470

   Connection Machine is a registered trademark of Thinking Machines Corporation.