840493n@aucs.UUCP (BILLY A. NICKERSON) (10/03/88)
Does anyone have any information on John von Neumann, specifically contributions to the field of computer science? Thanx in advance. Incidentally, references are okay too. Pleeze use e-mail. Bill N., The Beta-Ray Kid. ------------
jjb3281@acf3.NYU.EDU (Thanbo) (10/05/88)
I tried sending this by e-mail, but my mailer refused to cooperate.
A good book on the early history of computing is "The Computer from
Pascal to Von Neumann" by Hermann H. Goldstine (Princeton Univ Press,
1972). Goldstine worked with von Neumann, so it has some nice personal
color as well as lots of info. His basic ideas on computer architecture
are laid out in his "Preliminary Report on the Design of an Electronic
Computing Instrument", which can be found in various places, such
as von Neumann's collected papers, and perhaps more widely available
in "Computer Structure: Readings and Examples" by C. Gordon Bell and
Allen Newell.
As I understand it, his fundamental idea in computer architecture
was that -everything- should be serial, to minimize the complexity
of machinery. Down to bit-level serialism, using one-bit adders and
the like & circulating a word through the adder one bit at a time.
It was soon realized that bit-level parallelism was a Good Thing,
but we are stuck with his serial ideas otherwise, with one path
between memory and processing units, the so-called Von Neumann Bottleneck.
The bottleneck is seen since no matter how big the memory or how fast
the processor, processing speed is limited by the speed at which information
can travel between the CPU and the memory.
I hope this is some help.
Thanbo
jjb3281@acf3.nyu.edu
(Magnetic Drum Data Processing Machines Forever!!)