When I first learned
to program in 1964 we used an IBM 1620, fondly known
as CADET (Can't Add,
Doesn't Even Try). It had 20,000 memory cells, each
of which stored one
decimal digit. Words were delimited by a flag bit in
the
most significant
digit -- i.e. they were variable length. Since there were
no
integrated circuits
then, the whole desk-sized machine, complete with blinking
lights, console
switches, and a built-in IBM Executive electric typewriter
for
small amounts of
input and output, was made with individual transistors.
My
ex could type faster
than that machine, especially with a Selectric, but this
was before the
Selectric too.
For larger amounts
of input or output one of course used the read-punch unit.
You typed your
Fortran program on a keypunch, went to the machine, set the
switches, read in a
deck with the Fortran compiler, reset the switches, read in
your program and it
punched out the object code on a deck of cards. If you
needed any
subroutines (sine, exp, etc.) they were on a separate deck you
read
in and it punched
out the necessary code. Finally you took your compiled
deck
from the punch side,
reset the switches, and read it in and got your output
off
the
typewriter. If you had a lot of output, you punched it to a deck of cards
and
took them over to
the listing machine, put in an appropriate pegboard to get
the
formatting you
wanted, and printed out the deck.
The variable word
length was very useful for calculating Fibonacci numbers to
1000's of
digits. Incidentally, 3/5/08 last month is a Fibonacci date, the
last
one for about 5
years till 5/8/13. After that it's 8/13/21 and that's it for
this
century.
Please do not bend,
fold, staple or mutilate these cards.
Bob
