Shared Memory to Message Passing Rewrites (was RE: The use of a b eowulf.)

[Bohn Christopher A Capt AFRL/IFSD]

Good day,

> From: Szwedyk, Peter [mailto:Peter.Szwedyk@gs.com]
> Sent: Tuesday, June 22, 1999 2:22 PM
[...]
> It is my understanding that in a Distributed Shared Memory 
> system existing
> serial, multi-threaded apps would automatically (in most 
> cases) be able to
> take advantage of the cluster without a need for re-write.  
> However, DSM is
> currently not supported in a Beowulf system.

A quick semantic note (not intended as criticism, but rather to instigate a
different discussion, next paragraph) -- a multi-threaded app, while it may
run on a uniprocessor, is (by definition) not a serial application; it's a
concurrent application [1].

Which brings up a good question -- given that multithreaded apps already
have concurrent aspects worked out (though not necessarily optimally for
parallel/distributed computing), has anyone attempted to replace shared
memory constructs with message passing constructs in an existing
multithreaded app?

Take care,
cb

[1] The best distinction between "concurrent" and "parallel" that I've heard
is that concurrency implies doing two or more things at the same time (or
provide the illusion of that by using time slices), and parallelism implies
using concurrency on more than one processor.

*-*-*-*-*                                                     *-*-*-*-*
Christopher A. Bohn, Capt, USAF     || christopher.bohn@sn.wpafb.af.mil
Digital Simulation Systems Engineer || cbohn@computer.org
Collaborative Simulation Technology ||
          and Applications Branch   || v (937)255-4429x3576  (DSN785)
Information Directorate             || f (937)255-4511       (DSN785)
Wright Research Site                ||
Air Force Research Laboratory       || http://members.aol.com/EngrBohn/
         http://www.if.afrl.af.mil/div/IFS/IFSD/IFSD_home.html
*-*-*-*-*                                                     *-*-*-*-*