Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uflorida!gatech!hubcap!billwolf%hazel.cs.clemson.edu
From: billwolf%hazel.cs.clemson.edu@hubcap.clemson.edu (William Thomas Wolfe,2847,)
Newsgroups: comp.sw.components
Subject: Re: Linear Scalable S/W Components
Message-ID: <5447@hubcap.clemson.edu>
Date: 10 May 89 18:09:08 GMT
References: <166@oscsuna.osc.edu>
Sender: news@hubcap.clemson.edu
Reply-To: billwolf%hazel.cs.clemson.edu@hubcap.clemson.edu
Lines: 26

From article <166@oscsuna.osc.edu>, by stein@oscsuna.osc.edu (Rick 'Transputer' Stein):
> 1) In a multicomputer environment, one needs s/w components which
> are scalable.  Scalable in the sense that a particular component
> can be replicated (in a linear fashion concommitant with the #
> of processors in the system), and executed in a massively parallel
> environment.  [...] Do I get n-times single cpu performance
> when I run it on n processors (save for communication bound problems)?
> [...] One can not continue to create toxic waste dump software programs 
> and expect to have the engineering effort be reuseable without applying 
> knowledge of the target architecture.

    Absolutely.  The way I accomplish this is to write my components
    using Ada multitasking; the components can then be compiled on
    any desired system.  The Ada compiler for that system has knowledge
    of the target architecture, and generates code which maps the abstract
    model (independent processes communicating via the rendezvous mechanism)
    onto the number of physical processors actually available.

    Another approach to this problem which will be used in the future is
    the connection of large numbers of computer systems into a network
    such as the NECTAR system being implemented by Dr. H.T. Kung at
    Carnegie-Mellon; on such a system, processes can move to the system
    most appropriate for their execution.


    Bill Wolfe, wtwolfe@hubcap.clemson.edu