|
|
 |
 News
Seminar
On the Road to a Global Exascale Programming Paradigm
Serge G. Petiton
Laboratoire d'Informatique Fondamentale de Lille
Universite des Sciences et Technologies de Lille
Date
JEC 3117 - 2:30 p.m. to 3:30 p.m.
Refreshments at 2:00 p.m.
Abstract:
The Grid and High Performance Computing recent evolution leads to the
development of new innovative architectures. New computers will be built
as large, hierarchical, and heterogeneous architectures, integrating
grids and different programming paradigms to push forward the frontiers
of High Performance Computing. From Grid-like computing to the
optimization of multi-core processors and modern accelerators, the
end-users and software engineers will face new programming challenges.
They would require the use of novel adapted systems, methods, algorithms
and languages not previously available for existing supercomputing
systems. Other transversal topics as fault tolerance and electrical
power consumption optimization will be intrinsically associated with
Exascale computing. Expertise on all these different topics and their
interactions toward large integrated solutions are the sine qua non
conditions for any future efficient Exascale computing.
In this talk, we will propose a potential global programming paradigm
for future Exascale platforms, based on experimentations on grids,
clusters, multi-cores, accelerator programming for dense and sparse
linear algebra methods.
Founded on the YML framework and its high level graph description
language, we will explain how federate several existing programming
paradigms and how end-user expertise may be required at several levels
of systems, scheduling and basic programming. We will discuss multi-core
computing optimization, with respect to choose compress
formats. We will especially discuss the problems opened by such
objectives and how interdisciplinary researches have to be considered.
We will survey several research directions to contribute to this quest
toward the next computing frontier.
Bio:
Prof. Serge G. Petiton received his Ph.D. degree in 1988 and "Habilitation"
in 1993 from Pierre and Marie Curie University, Paris 6.
He was a post-doc student, registered at the graduate school, and an
associate research scientist at Yale University from 1989 to 1990.
He was a research fellow at the "Site Experimental en Hyperparallelisme"
from 1991 to 1994; and, several times during this period, he was a
visiting fellow at NASA-ICASE and at the AHPCRC in the United States.
Prof. Petiton leads the "Methodology and Algorithmic Programming" group
of the CNRS "Laboratoire d.Informatique Fondamentale de Lille", and he
actively participates in the INRIA "Grand Large" project. He is
director of the board of the ORAP Association to promote HPC and
participates in several French, European and International HPC
committees. He advised sixteen Ph.D. graduates and is currently
advising six. He has authored more than 80 articles in journals and
international conferences. Prof. Petiton led the French delegation on
"supercomputers" in Japan a few years ago and leads the HPC and Grid
computing axe of the Japanese-French Laboratory in Informatics.
Currently his main research interests are in Parallel and Distributed
Computing, Multi-Core and Hybrid Computing, High Performance Linear
Algebra, and Grid and Cloud Computing.
Hosted by: Center for Pervasive Computing and Networking (CPCN),
Scientific Computation Research Center (SCOREC)
Last updated: July 27, 2009
|
|
|
|
