[comp.parallel] Amdahls Law

middleto@UCSD.EDU (Tom Middleton) (04/27/88)

I have heard Amdahls law mentioned here and would like to quote it in a
paper I am writing.  Does anyone know someplace where it is published so
I can read it and give it proper credit in my references.

On a slightly related issue, has anyone used information from this network
in a paper and if so how do you give the poster proper credit.  (I assume
you just name the poster, his/her organization, and say UUCP.  Any better
way to do this?)

[ My guess is that it is "private communication" even though
  it is "public" Steve.
]
				   Thomas Earl Middleton
 G O				   aka T. Bear or T. Bare or Bare Bear
	P A D R E S !              UUCP  ...!ucsd!sdsu!middleto
                                   ARPA  middleto%sdsu.uucp@ucsd.edu

art%maui.berkeley.edu@RELAY.CS.NET (Arthur Goldberg) (05/03/88)

%A G. M. Amdahl
%T The Validity of the Single Processor Approach to Achieving Large
Scale Computing Capabilities
%I AFIPS Conference Proceedings
%V 30 
%D 1967
%X Well known (infamous ?) Amdahl's law that suggests that if x %
of an algorithm is not parallelizable then the maximum speedup is 1/x. 
Limits of vectorization. 

Arthur Goldberg
cs.ucla.edu