INSTITUTION NAME: Syracuse University

WORKPAGE TITLE: WebHLA

PROJECT TITLE: WebHLA based Metacomputing CMS for the Million Mines
Challenge

POC: David Bernholdt, bernholdt@npac.syr.edu, (315) 443-3857, fax
(315) 443-1973

CTA or PEI: Forces Modeling and Simulation (FMS)

PROJECT DESCRIPTION: 

In Year 3, we presented initial demonstrations of Parallel CMS at SC98
in Orlando, FL. The demonstrated system included Parallel CMS module
itself, running on Origin2000, as well as ModSAF module for vehicle
and terrain simulation and Mak Stealth Visualization module, running
on SGI workstations, with all machines located in the HPCMO booth at
SC98. We also have a distributed version operational, with Parallel
CMS running at Origin2000 platforms at the ARL or CEWES MSRCs, and
with other modules running on NPAC cluster. Test scenarios we were
running so far included heavy breach operation through a minefield of
some 30,000 mines located in Ft. Knox terrain. Ft, Belvoir, the
primary user and developer of CMS, is interested in simulations
including large scale minefields of 1 million and more mines, such as
used during the Gulf War or deployed in the Korea Demilitarized Zone.
We propose to address the “Million Mines CMS Challenge” in this
Year 4 project. For this purpose, we will need to enforce full
scalability of parallel CMS across a single Origin2000 system, as well
as to sustain a metacomputing operation of a set of such HPC platforms
or equivalent workstation/PC clusters. The Origin2000 scalability
problem is addressed in the associated Year 4 proposal, whereas here
we focus on the metacomputing coordination and integration. We will
use WebHLA infrastructure developed so far and we will build an
automated decomposer that distributes large minefields over a set of
machines so that each platform runs a Parallel CMS federate over a
suitable minefield component of subset, and all such components
cooperate with a Metacomputing Parallel CMS Federation. We intend to
use initially ARL and CEWES HPC facilities such as Origin2000 and NPAC
workstation and PC clusters. In the next step, we will include most
adequate HPC facilities from the selected Distributed Centers such as
SPAWAR or NRL. Initial demo of Metacomputing CMS will be presented
during SC99 where will try to attack for the first time the Million
Mines CMS Challenge.

PROJECT OBJECTIVES: 

Develop a Metacomputing CMS simulation running over multiple MSRCs and
capable to sustain distributed minefields of million or more active
mines.

DELIVERABLES: 

1) SC99 Demonstration of Metacomputing CMS with distributed minefield
   components running at ARL, CEWES and NPAC sites; 

2) Sustained support for large scale distributed minefields with
   components in ARL, CEWES, NPAC and selected Distributed Centers such
   as SPAWAR and NRL.

CUSTOMERS/END USERS: 

1) Night Vision Lab for the countermine R&D; 2) HPCMO for general
publicity purposes; 3) Possibly JSIMS for realistic modern mine
warfare simulations.

BENEFIT TO THE WARFIGHTER: 

HPC FMS technology demonstration will help marketing the unique HPC
capabilities and will facilitate HPC technology insertion into the
advanced joint simulation systems and new generation Virtual
Prototyping and Simulation Based Acquisition systems.

PROJECT DEPENDENCIES AND SCOPE: 

This project depends on the completion of the “SPEEDES + HLA Based
Scalable Parallel CMS Federate” project.

RISK ELEMENT:

 1) User community for Parallel CMS is not fully recognized yet; 2)
Networking challenges of MBONE (ModSAF DIS) + Metacomputing (Parallel
CMS) integration are yet to be understood and addressed. 3) It may be
necessary for ARL and other participating sites to acquire licenses
for some software packages. 4) In all metacomputing activities, there
are significant issues around access/security and resource scheduling
that need to be addressed, either on a one-time fashion, or in
general. Such issues are beyond the scope of this proposal.

REQUIRED FUNDING LEVEL: