Hardware Reference
In-Depth Information
In 1964 a tiny, unknown company, Control Data Corporation (CDC), intro-
duced the 6600, a machine that was nearly an order of magnitude faster than the
mighty 7094 and every other machine in existence at the time. It was love at first
sight among the number crunchers, and CDC was launched on its way to success.
The secret to its speed, and the reason it was so much faster than the 7094, was that
inside the CPU was a highly parallel machine. It had several functional units for
doing addition, others for doing multiplication, and still another for division, and
all of them could run in parallel. Although getting the most out of it required care-
ful programming, with some work it was possible to have 10 instructions being ex-
ecuted at once.
As if this was not enough, the 6600 had a number of little computers inside to
help it, sort of like Snow White and the Seven Vertically Challenged People. This
meant that the CPU could spend all its time crunching numbers, leaving all the de-
tails of job management and input/output to the smaller computers. In retrospect,
the 6600 was decades ahead of its time. Many of the key ideas found in modern
computers can be traced directly back to the 6600.
The designer of the 6600, Seymour Cray, was a legendary figure, in the same
league as Von Neumann. He devoted his entire life to building faster and faster
machines, now called
supercomputers
, including the 6600, 7600, and Cray-1. He
also invented a now-famous algorithm for buying cars: you go to the dealer closest
to your house, point to the car closest to the door, and say: ''I'll take that one.''
This algorithm wastes the least time on unimportant things (like buying cars) to
leave you the maximum time for doing important things (like designing supercom-
puters).
There were many other computers in this era, but one stands out for quite a dif-
ferent reason and is worth mentioning: the Burroughs B5000. The designers of
machines like the PDP-1, 7094, and 6600 were all totally preoccupied with the
hardware, making it either cheap (DEC) or fast (IBM and CDC). Software was al-
most completely irrelevant. The B5000 designers took a different tack. They built
a machine specifically with the intention of having it programmed in Algol 60, a
forerunner of C and Java, and included many features in the hardware to ease the
compiler's task. The idea that software also counted was born. Unfortunately it
was forgotten almost immediately.
The invention of the silicon integrated circuit by Jack Kilby and Robert Noyce
(working independently) in 1958 allowed dozens of transistors to be put on a single
chip. This packaging made it possible to build computers that were smaller, faster,
and cheaper than their transistorized predecessors. Some of the more significant
computers from this generation are described below.
By 1964 IBM was the leading computer company and had a big problem with
its two highly successful and profitable machines, the 7094 and the 1401: they
