>Shawn T. Rutledge wrote:
> Interesting article here:
> http://www.ddj.com/columns/history/2001/0106hc/0106hc001.htm
Jerome Fine replies:
Figure 3 sure looks similar to what I remember of an IBM 650 that
I used in 1960, some ten years later. The reader/punch was still the
same. I also feel that the actual IBM 650 unit was similar in looks
on the outside to the IBM 604 shown. But inside, there must have been
a huge difference. The IBM 650 allowed 3 drums (if I remember
correctly - or maybe one big one that was 3 times the size of the
minimum variety). In any case, the system I first cut my real computing
teeth on (I had had some acquaintance with computers about two years
prior when I worked for an insurance company during the summer)
was an IBM 650 unit with 6000 words (10 decimal digits each) which
could represent either one instruction or one integer/floating point
number. The address range for the words was 0000 to 5999 and
the op code in each instruction was the first two decimal digits (no
new fangled binary in an IBM 650 at this point). Two operands
followed with the first being a data value and the second operand
always being the address of the next instruction. SOAP (Symbolic
Optimal Assembly Program) was used to place the next instruction
on the drum in a location based on how long the operation took
and where the drum would be in that time. Obviously, the experienced
programmer place all inner loops at the very start of the program with
both initiation code and termination code coming at the end when
it did not matter if all the best locations had been used.
The IBM 650 I work with also had 60 words (address 9000 to 9059)
of directly addressable storage - I guess in vacuum tube circuits. These
were almost always used for data with only the smallest of inner loops
being place there for speed.
Both input and output for the IBM 650 was punched cards - a 704 was
used to list the SOAP output at one instruction per card and another
program converted that deck of cards to more than one instruction
per card - either 3 or 5 or 7 if I remember correctly - with the
maximum being 7 (since 7 * 10 = 70 and 4 columns were reserved for
for the load address) and each card in the program deck was numbered
in the last 4 columns - I think
In any case, as slow as the IBM 650 was by today's standards, it was
very fast compared to a mechanical calculating machine that took seconds
to do a multiply and tens of seconds to do a division if that were even
possible - could a Frieden do a division?
What this reply is about is that even ten years later (1960 as opposed to
the depicted systems in 1948), the pace of change was still slow and
IBM was still THE IBM and the seven dwarfs were just starting to
poke up their collective heads. Very little changed between 1950 to
1960 except that the stage was being set for the changes that were to
come by 1970 by which time transistors were totally dominant and
mass??? storage had arrived in the form of core memory measured
in KBytes. The growth seems to have been exponential ever since.
Now that was just 40 years ago. What will computers be like in 2040?
Those youngsters whose first computer was a Pentium can't even begin
to realize how far commercial computing has gone in just 50 years.
While the HAL in 2001 may be another 50 years or even a hundred
years, eventually there will be programs that will begin to rival the lower
level of normal human administrative functions which do not require any
imagination and which a canned program will be able to manage in a
competent manner. But this last paragraph is the future and OT.
Sincerely yours,
Jerome Fine
Received on Mon Jul 02 2001 - 19:57:49 BST