Electronica

Microprocessolc

Nick Tredennick

The del elopment of the
computer and the evolution
of t he integrated circuit have

been intertwined since the
first commercial I C appeared

hen Univac delivered its first commercial computer in 1951,
it set off a chain of events that continues to unfold. Although
the Univac I used vacuum tubes for the switching elements,
the transistor-having beeninvented in 1947-soon displaced the vacuum tube as the basic switching element in digital designs. After the
transistor’s introduction, computers improved rapidly, becoming smaller,
faster, more capable, and more reliable.
The first practical IC was fabricated in 1959 at Fairchild and at Texas
Instruments. Its commercial introductionin 1961by Fairchild further sped
improvements in computerimplementation. Improvements in semiconductor fabrication made the computer faster, more functional, and
cheaper, too, lowering barriers to entry for applications and for manufacturers alike. Mainframe manufacturers were soonjoined by a growing
number of minicomputer manufacturers, and the computer business was
off and running.
At the time of its introduction, the computer was viewed as a fastcalculating engine. But the events it precipitated had dramatic consequences
because it was fundamentally a breakthrough for problem solving. Before
the computer, most problems were solved by direct implementation: You
built specific hardware to solve a specific problem. Combinational logic
implemented both the algorithm and the computational resources.
The size of the problem that could besolved was limited by money: You
couldn’t solve a problem if you couldn’t afford the hardware.
Separating the algorithm from the resources, the computer implemented the problem-solvingalgorithm in cheap, plentiful memory (in a
hierarchy that might include magnetic core, magnetic disk, magnetic
tape, paper tape, and punched cards). It implemented the computational
resources in scarce, expensivehardware. The computer exploited temporal and spatial locality properties of the problem-solving method to
gain performance close to that of the expensiveresource (hardware computing elements) at a cost close to that of the cheap resource (memory).
The computer iterated to solve a problem. This meant the computer
could solve problems too large to be affordable with combinational
resources-ifyou were willing to wait for t he answer. The computer traded
time for hardware resources, allowing the computer to solve a larger range
of problems than could be affordably tackled with combinational solutions. In addition, the algorithm sat in memory. This memory could be
readily altered, so t he cost of the computer could be amortized across a
range of applications, which made it a mostattractive tool.

in 1961. The author explores
microprocessor history and
ponders future developments.
0018-9162/96/85.00 0 1996 IEEE

GROWTH OF A BUSINESS
Meanwhile, the semiconductor business was booming as industrial
and commercial products converted from vacuum tube to semiconductor implementations. RTL (resistor-transistor logic) components,
October 1996

introduced in 1961,weresoon superseded by TTL (transistor-transistor logic) components. The family of TTL
parts grew in popularity and variety as semiconductor
fabrication processes improved. Because it was a family
of electrically compatible logic macro functions (rather
than of individual transistors), TTL raised the efficiency
of design engineers. This served to increase the rate of
innovation in computerimplementations. At first, TTL
components displaced small groups of individual transistors. Later, TTL MSI (medium-scale integration) and
LSI (large-scale integration) parts displaced small
groups of TTL SSI (small-scale integration) parts. As the logic macro functions
b ecame larger and more complex, the
products they were in became smaller,
ogether with
cheaper, and more reliable.
i ts "killer...