Newsgroups: comp.lang.functional
Path: utzoo!utgpu!watserv1!maytag!watdragon!abrodnik
From: abrodnik@watdragon.waterloo.edu (Andrej Brodnik (Andy))
Subject: Re: Program transformations for parallelism
Message-ID: <1990Dec3.063533.27464@watdragon.waterloo.edu>
Summary: History ...
Organization: University of Waterloo
References: <1824@xenon.kcl-cs.UUCP> <3100@sequent.cs.qmw.ac.uk>
Date: Mon, 3 Dec 90 06:35:33 GMT
Lines: 14

In article <3100@sequent.cs.qmw.ac.uk> mmh@cs.qmw.ac.uk (Matthew Huntbach) writes:
>Many algorithms are inherently sequential, and thus attempts
>to parallelise programs based on them have proved disappointing.
>I think a better approach is to develop new parallel
>algorithms and implement them in your new parallel language.


And what to do with already written programmes? Should we employ a whole
mankind to convert their algorithms into more appropriate form? I think that
the main reason why to parallelise programmes lies in "history".

Andrej
 



Brought to you by Super Global Mega Corp .com