Numerical simulation is playing an increasingly important role in natural sciences and in many areas of engineering. This development is mainly attributed to the performance capabilities of modern computers. By now peak performance is dimensioned in PetaFlop/s (quadrillion floating point operations per second) and available memory is given in TeraBytes (trillionen bytes). Such performance values can only be achieved by parallel processing of data.
The implementation of numerical methods on parallel platforms and supercomputers requires development of specific numerical approaches. Investigation of corresponding methods and techniques is the objective of this seminar. By means of problem statements different aspects of the solution process on parallel machines shall be discussed in detail. If requested benchmark tests can be performed on the high performance computers at Karlsruhe Institute of Technology or on multi-GPU-systems.
Interested candidates are kindly requested to send an email to firstname.lastname@example.org. It is expected to give a beamer presentation and to deliver a written summary of the talk. Knowledge in numerical mathematics and programming is desired. Topics of this seminar may result in master or diploma theses.
First meeting: Thursday, July 15, 2010, 16:00h, Room 256, Build. 01.86, 2nd floor