The Grid Computing Research
Group consists of faculty, graduate and
undergraduate students in the Faculty of Computer Science at the
University of New Brunswick who are working in the area of grid
computing. Computational grids offer seamless access to distributed
computational resources, and incorporate heterogeneous hardware,
operating systems, and system configurations. The present focus of our
group is twofold: mesh-based high performance computing applications
distributed over multiple parallel computers, and coupling of
high-performance Computer Aided Engineering software with Model
Predictive Control software. We have developed PaGrid, a mesh
partitioner for computational grids, which aims to minimize the
estimated execution time of the application, and are presently
developing grid-based model-predictive control (GridMPC). Ongoing work
includes improvements to PaGrid, testing with real applications, and
further development of GridMPC. We use the computing facilities in the
Advanced Computational Research Laboratory (ACRL).
Faculty
Graduate Students
- More on their way in Sept. 2008!
Alumni
- Basile Clout, MCS thesis: A Test-Bed Framework for
Heterogeneous Mesh Partitioners, May 2008 (supervisor: Eric Aubanel)
- Kambiz Ghazinour Naini, MCS thesis: A
Linear Solver for Benchmarking
Partitioners, November 2007 (supervisors: Eric Aubanel and
Ruth Shaw)
- Xiaochen Wu, MCS
thesis: Flexible Mesh
Partitioning for
Computational Grids, April
2007 (supervisor: Eric Aubanel)
- Shuangyu Yu, MCS thesis: PaGridT:
A Mesh Partitioner with Tabu Search for Computational Grids,
December 2006 (supervisor: Eric Aubanel)
- Benoit Philipps, MCS thesis: A
Grid Architecture for the Remote Control of Industrial Systems,
December 2005 (supervisor: Eric Aubanel)
- Tanvire Elahi, MCS thesis: Efficient
Solutions of Systems of
Equations with Banded Coefficient Matrices and Their Implementation as
a Grid Service, August 2005
(Co-supervised by Eric Aubanel and Ruth
Shaw)
- Renaud Wanschoor, MCS
thesis: Mesh Partitioning for
Computational Grids, October
2004 (supervisor: Eric Aubanel)
- Sili Huang, MCS thesis: PaGrid:
A Mesh Partitioner for
Computational Grids, April 2003
(supervisors: Eric Aubanel and
Virendra Bhavsar).
- PaGrid, a mesh partitioner
for computational grids
Mesh-based applications, such
as those that involve the numerical
solution of partial differential equations, may be able to take
advantage of the performance of computational grids. To do so will
require mesh partitioners that take the heterogeneity of the
computational platform into account. We have developped Pagrid, a
heterogeneous mesh partitioner which balances the estimated execution
time of processors. The code is available at Google.
- Benchmarks for graph partitioners
LiSBench
(Linear Solver for Benchmarking) is a benchmark that enables comparison
of graph partitioners. It solves the 2D Laplace equation using
triangular finite elements and the conjugate gradient method, on a
partitioned mesh provided by the user. It also serves to motivate
further work on developing benchmarks for graph partitioners. The
code is available at Google.
- GridMPC
A number of industrial
applications are now automatically
controlled by computers. Model Predictive Control (MPC) software
controlling these applications needs to have a knowledge of how the
process reacts. Obtaining these data by running experiments is
technically difficult and highly time-consuming. However, these
experiments can be replaced by computer simulations. It has been proved
that MPC software controlling a machine in closed-loop mode can be
coupled with Computer Aided Engineering (CAE) software running
simulations in open loop mode. This raises some issues, mainly because
the required computing resources are not often available locally, and a
solution has to be found to have this idea implemented and working in
real time.
This is why the Grid Computing Research Group and the Department of
Mechanical Engineering of UNB are trying to set up a grid architecture
to use recent distributed computing techniques to efficiently couple
CAE software and model predictive control software controlling an
industrial process.
Profs. Aubanel and Bhavsar organize an annual workshop on high
performance grid computing, in collaboration with Michael Frumkin from
Google.
- Fifth
High-Performance Grid Computing Workshop (HPGC '08), in conjunction
with IPDPS 2008, April14, 2008, Miami, Florida.
- HPGC '07,
in conjunction
with IPDPS 2007, March
26 , 2007,
Long Beach, California.
- HPGC '06,
in conjunction
with IPDPS 2006, April 29, 2006, Rhodes Island, Greece.
- HPGC
'05
(joint workshop with High-Level Parallel Programming Models), in
conjunction with IPDPS
2005, April 4 2005, Denver, Colorado.
- HPGC
'04, in
conjunction with IPDPS 2004, April 30 2004, Santa Fe, New Mexico.
- E. Aubanel, "Resource-Aware
Load Balancing of Parallel Applications", accepted by the Encyclopedia of
Grid Computing, February 2008.
- K. Ghazinour, R.E. Shaw, E. Aubanel, and L.E. Garey, "A Linear
Solver
for Benchmarking Partitioners", Workshop on Parallel and Distributed
Scientific
and Engineering Computing (PDSEC), in Proc. 22nd Intl. Parallel and
Distributed Processing Symposium, April 2008, Miami, Florida,
IEEE CD-ROM (2008), 8 pages.
- E. Aubanel and X. Wu,
"Incorporating Latency in Heterogeneous
Graph Partitioning", Workshop on Parallel and Distributed Scientific
and Engineering Computing (PDSEC), in Proc. 21st Intl. Parallel and
Distributed Processing Symposium, March 2007, Long Beach California,
IEEE CD-ROM (2007), 8 pages.
- B. Philipps, E. Aubanel, A.
Healy and A. Gerber, "GridMPC:
a
Service-Oriented Grid Architecture for Coupling Simulation and Control
of Industrial Systems", Workshop on Service Oriented Techniques
(SOT06), at the Eighth International Conference of E-Commerce (ICEC06),
Canadian National Research Council, in press, 8 pages. An expanded
version appears in the following:
- B. Philipps, E. Aubanel, A. Healy and A. Gerber, "A
Service-Oriented Architecture for Coupling Simulation and Control
of Industrial Systems", Int. J. Business Process Integration and
Management, in press, 2008, 10 pages.
- S. Huang, E. Aubanel, and
V.C. Bhavsar, "PaGrid: A Mesh
Partitioner for Computational Grids", Journal of Grid
Computing, Vol.
4 No. 1, pp. 71 - 88, 2006.
- R. Wanschoor and E. Aubanel,
“Partitioning and Mapping of
Mesh-Based Applications onto Computational Grids”, 5th
IEEE/ACM
International Workshop on Grid Computing (Grid 2004), Nov. 8, 2004,
IEEE Computer Society, to appear Nov. 2004.
- R. Wanschoor and E. Aubanel, Mesh Partitioning for
Computational Grids, Second
Annual Conference on Communication
Networks and Services (CNSR-2004), May 19-21, 2004, IEEE Computer
Society, pp. 325-328.
- W. Huang, E. Aubanel, and
V.C. Bhavsar, Mesh
Partitioners for
Computational Grids: A Comparison,
Proc. of the 2003 International
Conference on Computational Science and Its applications (ICCSA 2003),
Montreal, PQ, Canada, May 18-21, 2003, Lecture Notes in Computer
Science 2267-2269, Springer Verlag (2003), pp. 60-68.
- Partial financial support
provided by:
Revised: June 3, 2008 by Eric Aubanel
Grid
Computing Research Group