The PVM Project

   A. Beguelin, J.J. Dongarra, G.A. Geist, W. Jiang, 
      R. Manchek, K. Moore, and V.S. Sunderam
 
Description
-----------
PVM (Parallel Virtual Machine) is a software system that enables
a collection of heterogeneous computers to be used as a coherent and
flexible concurrent computational resource.  The individual 
computers may be shared- or local-memory multiprocessors, vector 
supercomputers, specialized graphics engines, or scalar workstations, 
that may be interconnected by a variety of networks, such as ethernet, 
FDDI, etc.  PVM support software executes on each machine in a 
user-configurable pool, and presents a unified, general, and powerful 
computational environment for concurrent applications.  User programs 
written in C or Fortran, are provided access to PVM through the use 
of calls to PVM library routines for functions such as process initiation, 
message transmission and reception, and synchronization via barriers 
or rendezvous.  Users may optionally control the execution location of 
specific application components; the PVM system transparently handles
message routing, data conversion for incompatible architectures, and 
other tasks that are necessary for operation in a heterogeneous, network 
environment.

Applications and Environments
------------ --- ------------
PVM is ideally suited for concurrent applications composed of many 
interrelated parts.  PVM is particularly effective for heterogeneous
applications that exploit specific strengths of individual machines 
on a network.  As a loosely coupled concurrent supercomputer environment 
PVM is a viable scientific computing platform.  PVM system has been used 
for a number of applications such as molecular dynamics simulations, 
superconductivity studies, distributed fractal computations, matrix algorithms,
and in the classroom as the basis for teaching concurrent computing.

Status and Availability
------ --- ------------
The beta release of PVM 3.1 is now available and may be obtained by sending
electronic mail to  netlib@ornl.gov  with the message
      send index from pvm3
There are a number of improvements planned for the beta version.
A notable one is performance; we expect to improve the performance over 
the course of the next few months as we tune and adjust the internals.
The beta release contains a users guide and installation notes;
PVM has been tested on a wide variety of systems.

Some of the new features of PVM include:
o) Runs on nodes of multiprocessors - initially Intel iPSC/860
      other machines like CM-5, Paragon to be added later.

o) Dynamic configuration - computers can be dynamically added and
      deleted from the parallel virtual machine by the application
      or manually.

o) Multiple message buffers - allowing easier development of 
      PVM math libraries, graphical interfaces, etc.

o) Signals - PVM tasks can send signals or notify sets of tasks
      about events.

o) Dynamic process groups - user-defined process grouping
      for functions like broadcast and barrier.
      groups can overlap and change dynamically during an application.

o) Cleaner naming convention for PVM routines - C routines start with pvm_
      Fortran routines start with pvmf. For example pvm_send.

o) Improved routines - receive by source or type (or user defined context)
      spawn multiple tasks with options for argv and debug,
      pack and unpack messages using stride, and
      more query functions.

o) New more comprehensive User's Guide and PVM 3 Reference Manual.

PVM is an ongoing research project, as such support at levels appropriate for 
a research project may be reasonably expected.  Questions and comments can 
be sent to  pvm@msr.epm.ornl.gov.

This project is supported in part by the Applied Mathematical Sciences 
subprogram of the Office of Energy Research, U.S.  Department of Energy,
by the National Science Foundation Science and Technology Center (CRPC),
and by the Science Alliance a state supported program
at the University of Tennessee.