Parallel Computing and Numerics

Prévu en mars 2013

place : Université de Strasbourg
exact location will be announced later
Organizers : V. Heuveline, S. Ritterbusch (KIT Karlsruhe), P. Helluy, C. Prud’homme, P. Clauss (IRMIA Strasbourg)
time : 
exact date will be announced later
Audience : the lecture is open to scientific master students. The registration is mandatory and can be done at
Students who are not from Strasbourg or Karlsruhe have to send a CV to the organizers before being allowed to register.
lecture is held in English
Course Description
Modern supercomputers and the development of adapted parallel algorithms facilitate the solution of problems within a few hours that would be unsolvable on desktop computers or would take weeks or months. Thus, simulation of complex phenomenons in natural and engineering sciences and corresponding parameter studies are enabled.
The main topics of the lecture are :
Mathematical modelling and domain decomposition methods for the solution of PDEs
Hardware architectures and parallel programming paradigms for shared and distributed memory systems
Parallel data structures and algorithms
Methods for assessment of parallel performance
Recent developments in multi-core technologies (e.g. GPU, FPGA, Cell, Manycore)

After successfully completing this course, participants will be able to :
Implement linear algebra routines on parallel computers by means of MPI, OpenMP and CUDA/OpenCL
Solve problems related to partial differential equations on parallel computers
Handle parallel computing platforms, paradigms and libraries
Have good knowledge in parallel solution techniques for linear systems of equations
Report on theoretical and practical experience on parallel computers
Conduct performance analysis and scalability assessment

Course materials will be provided during the lecture.

T. Mattson et al : Patterns for Parallel Programming. Addisson- Wesley, 2004
C. Lin : Principles of Parallel Programming. Pearson, 2009
C. Breshears : The Art of Concurrency. Reilly, 2009
J. Schüle : Paralleles Rechnen. Oldenbourg 2010
G. Alefeld et. al. : Parallele numerische Verfahren. Springer, 2002
W. Huber : Paralleles Rechnen. Oldenbourg, 1997
H. Schwandt : Parallele Numerik. Teubner, 2003
T. Rauber et al. : Parallele Programmierung. Springer, 2007

Necessary requirements are basic knowledge in programming languages C or C++. The required knowledge in numerical mathematics and treatment of partial differential equations can be acquired within the course.