Newsgroups: comp.parallel.pvm
From: andy@henry.jpl.nasa.gov (Andy Boden)
Subject: Re: PVM on PGON
Organization: Jet Propulsion Laboratory (California Institute of Technology)
Date: 27 Sep 1994 19:39:05 GMT
Message-ID: <369scp$s78@elroy.jpl.nasa.gov>

In article <369nblINNi1i@CS.UTK.EDU>, wjiang@thud.cs.utk.edu (weicheng jiang   ) writes:
> In article <369euu$2pbr@info2.rus.uni-stuttgart.de>, zra60100@nfhsg2.rus.uni-stuttgart.de (Michael Resch) writes:
> |> hi,
> |> 
> |> a question that is obviously of importance for our users here is,
> |> whether PVM is able to communicate outside a paragon e.g. to a
> |> workstation to allow distributed computation in a supercomputing
> |> environment were visualisation is done on graphics workstation
> |> and computation is done on the paragon.
> |> 
> |> has someone out there in netspace ever seen a solution of this problem?
> |> 
> |> thanks for any help
> |> 
> 
> You should be able to do that now. I've tried it on our Paragon.
> 
> -Weicheng Jiang
> University of Tennessee

Request for follow-up information please.  What *exactly* have you
been able to do on your Paragon?  Configure a virtual machine with the
paragon nodes and other unix boxes simultaneously?  If so:

1) Is it possible to start the PVM machine from the unix box, or must
the unix boxes be added from the paragon side.

2) Is it possible for a process on the unix side to spawn processes on
the paragon side (presumably on the worker nodes)?  If not, is the
converse true (spawn on unix from paragon)?

The last time I heard you speak to this issue you said something like
"maybe someday, but not now" (paraphrase).  So, has the time come
where the paragon and other MPP boxes can indeed be integrated into
the PVM heterogeneous environment model?

							- A.B.

-- 


"...it doesn't matter how beautiful your theory is, it doesn't matter
how smart you are -- if it doesn't agree with experiment, it's wrong."

						- R.P. Feynman

