Source : Free On-Line Dictionary of Computing
IBM 1620
A computer built by {IBM} and released in late
1959. The 1620 cost from around $85,000(?) up to hundreds of
thousands of dollars(?) according to the configuration. It
was billed as a "small scientific computer" to distinguish it
from the business-oriented {IBM 1401}. It was regarded as
inexpensive, and many schools started out with one.
It was either developed for the US Navy to teach computing, or
as a replacement for the very successful {IBM 650} which did
quite well in the low end scientific market. Rumour has it
that the Navy called this computer the CADET - Can't Add,
Doesn't Even Try.
The {ALU} used lookup tables to add, subtract and multiply but
it could do address increments and the like without the
tables. You could change the number base by adjusting the
tables, which were input during the boot sequence from
{Hollerith} cards. The divide instruction required additional
hardware, as did {floating point} operations.
The basic machine had 20,000 decimal digits of {ferrite core
memory} arranged as a 100 by 100 array of 12-bit locations,
each holding two digits. Each digit was stored as four
numeric bits, one flag bit and one parity bit. The numeric
bits stored a decimal digit (values above nine were illegal).
Memory was logically divided into fields. On the high-order
digit of a field the flag bit indicated the end of the field.
On the low-order digit it indicated a negative number. A flag
bit on the low order of the address indicated {indirect
addressing} if you had that option installed. A few "illegal"
bit combinations were used to store things like record marks
and "numeric blanks".
On a {subroutine} call it stored the {return address} in the
five digits just before the entry point to the routine, so you
had to build your own {stack} to do {recursion}.
The enclosure was grey, and the core was about four or five
inches across. The core memory was kept cool inside a
temperature-controlled box. The machine took a few minutes to
warm up after power on before you could use it. If it got too
hot there was a thermal cut-out switch that would shut it
down.
Memory could be expanded up to 100,000 digits in a second
cabinet. The cheapest package used {paper tape} for I/O. You
could also get {punched cards} and later models could be
hooked up to a 1311 {disk drive} (a two-{megabyte} {washing
machine}), a 1627 {plotter}, and a 1443 {line printer}.
Because the 1620 was popular with colleges, IBM ran a clearing
house of software for a nominal cost such as {Snobol},
{COBOL}, chess games, etc.
The model II, released about three years later, could add and
subtract without tables. The {clock period} decreased from 20
to 10 microseconds, instruction fetch sped up by a few cycles
and it added {index registers} of some sort. Some of the
model I's options were standard on the model II, like
{indirect addressing} and the {console} {teletype} changed
from a model C to a {Selectric}. Later still, IBM marketed
the {IBM 1710}.
A favorite use was to tune a FM radio to pick up the
"interference" from the lights on the console. With the right
delay loops you could generate musical notes. Hackers wrote
{interpreters} that played music from notation like "C44".
1620 consoles were used as props to represent {Colossus} in
the film "The Forbin Project", though most of the machines had
been scrapped by the time the film was made.
{A fully configured 1620
(http://uranus.ee.auth.gr/TMTh/exhibit.htm)}.
{IBM 1620 console picture
(http://www.foldoc.org/pub/IBM1620-console.jpg)}.
{IBM 1620 at Tuck School of Business, Dartmouth College,
Hanover, NH, USA
(http://www.foldoc.org/pub/IBM1620-Tuck1960s.jpg)} (Thanks
Victor E. McGee, pictured).
["Basic Programming Concepts and the IBM 1620 Computer",
Leeson and Dimitry, Holt, Rinehart and Winston, 1962].
(1997-08-05)