From: tony@CS.MsState.Edu (Tony Skjellum)
Newsgroups: comp.parallel.pvm
Subject: PVM interoperability with MPI (at various levels)
Date: 26 Sep 1998 04:09:09 GMT
Organization: Mississippi State University
Message-Id: <6uhpd5$a1r$1@nntp.msstate.edu>
Summary: Unify II is a reworked tool for mixing MPI and PVM
Keywords: PVM, MPI, interoperation


Dear PVM world:

One of the Mississippi State University MPI implementations, besides
ANL/MSU MPICH, was the Unify subset, which patched PVM 3.x to support
both PVM and MPI.  This was done in 1995-96 period, and we still get
downloads.  I am trying to gauge continued interest in this capability,
and get expressions of interest of further efforts in these areas:

* update support for Unify as Unify II with more features
* support transition from PVM to MPI
* support moving codes between PVM and MPI
* does anyone have a compelling need for such conversions?
* does the Unify library approach help people?
[by the way, it is still accessible at ftp.erc.msstate.edu]
* does anyone using PVM miss overlapped communication and computation?

I am writing to make sure we make appropriate future investment in Unify.
We also have some new ideas that we might do with this, if people would use
it.

I would like to know if people have done interprocedural analysis programs
to convert PVM codes to MPI or vice versa.

To save bandwidth, please post responses to me, and after responses die
down, I will post one complete response.  Please include web-accessible
or journal/conference article references, if referring to systems or tools
other than the well known ones like MPICH and PVM.

Thanks,
Tony



-- 
A. Skjellum, PhD, Assoc. Prof. of Computer Science; Mississippi State Univ.
http://www.cs.msstate.edu/~tony; tony@cs.msstate.edu; 601-325-8435 (FAX -8997)
"Mississippi (n): small US state where opportunities abound." [Try MPI/RT!]