Forecast: computer revolution;Forecast: Supercomputers;

              /MH^sga^z?3
computer Revolution
We are now in the midst of a revolution of the same magnitude and power as
the industrial revolution of the nineteenth century. It is changing our society,
our skills, and the character of employment in the United States. This revolu­tion
is driven by advances in microelectronics, transforming our contem­porary
world from an industrial to an information society. • At the heart of
this revolution is the computer—destined to become an essential and pervasive
tool in the 1980s, as the calculator was in the 1970s. And with this will come a
profound change in our understanding of what a computer is, and what it can
do. • The computer of the 1980s will adapt much more closely to the way the
human mind works, rather than forcing us to adapt to the computer's way of
operating. It will be easier to use, and will take minutes—rather than hours—to
learn. Technical advances will allow better graphics, greater integration of
software packages, better communication between computers, and input and
output devices, such as the "mouse," that allow users to work with the
computer much more effectively. • Though their precise nature cannot be
predicted, these changes will impact across the entire spectrum of American life.
STEVEN JOBS
Chairman
Apple Computer, Inc.
supercomputers
Supercomputers today are widely used for engineering design, oil exploration,
medical diagnosis, weather forecasting military development, and scientific
research. In many areas, much faster computers are required, however.
Advances in semiconductor technology (very-large-scale integrated circuits)
and computer architecture (highly concurrent processing) will lead to much
faster systems in this decade. • Today's supercomputers are organized in a
rather rigid manner that allows them to deliver several hundred million
numerical operations—adding or multiplying for example—per second on
arrays of data at peak speeds. However, in many applications their efficiency
is about 10 percent, making them much less cost-effective than they could be.
Current research in multiprocessing (as opposed to traditional array
processing) is demonstrating how to improve supercomputer efficiency. New
architectures together with powerful software systems should lead to super­computers
that deliver over a billion numerical operations per second and
have much higher peak performances.
DAVID KUCK
Department of Computer Science
University of Illinois
Champaign- Urbana