Quote:
Originally Posted by madbadgalaxyman
We learn in IT class that the stored program computer in common use has the "von Neumann" architecture.
Of course this could eventually end up as being one of those "oft repeated attributions" that ends up being proved wrong after further historical analysis......
|
Hi Robert,
Thanks for the post.
It proved to be an accidental and unfortunate attribution.
The successor to ENIAC was to be EDVAC and Von Neumann came to the
project by chance.
ENIAC had used an awkward plug-board to program it. When Mauchly
later developed the mercury delay line, it became practical in an
engineering sense to afford the stored-program paradigm and
it became incorporated into the plans for EDVAC.
Von Neumann volunteered to draft an internal memorandum documenting
EDVAC's design and Eckert, Mauchly and the team accepted, grateful
to have him on-board.
Unfortunately Goldstine, who was administrator of the project, more
widely circulated this incomplete draft report which only had Von Neumann's
name on the cover and attributions were to be added later.
When Von Neumann left the ENIAC group along with Goldstine
to work on the Institute of Advanced Studies (IAS) machine in the
town of Princeton, they further elaborated the stored-program
concept. This is why Von-Neumann architecture machines are also
often referred to as Princeton architecture machines.
The stored-program paradigm lends itself to the idea of self-modifying
code and this was elaborated in several reports by the IAS group.
However, they mainly restricted themselves to the idea of using it
for a computed address modification similar to what an index register
does on some machines.
The concept of a linear address space and a Program Counter would
all flow out of this work.
I've often wondered if Von Neumann had lived longer whether he would
have formulated a computer that exploited the self-modifying code
concept much further in some weird and wonderful way, much akin
to his work on self-reproducing automata.
Toward the end of his life he wrote a beautiful little book called "The
Computer and the Brain". There use to be a copy in the UNSW library
which I read decades ago.
Getting back to the mathematics and war theme, Von Neumann's
interest in ENIAC was not only academic but for him of pressing
importance. He was a major part of the brain power behind the
implosion device at Los Alamos and designing and computing
the spherical charges to uniformly implode the plutonium sphere
was a incredibly challenging problem. Von Neumann brought to
the project the concept of solving the nonlinear partial-differential
equations numerically and they were employing rooms full of
woman with Marchant calculators crunching the numbers.
When he had a chance meeting with Goldstine at a railway station
and Goldstine told him he was working on ENIAC, he instantly
recognized the utility it had to the problem at hand. A problem that
could bring about a quick end to the war.
