Particle-Mesh-Methods on Next Generation Supercomputers
Particle-Mesh-Methods on Next Generation Supercomputers
Bussmann, M.
Abstract
In the coming years supercomputers will have millions of compute cores.
It is estimated that the computing power will grow by a factor of 1000 in the coming years while the memory will only grow by a factor of 10.
This means that at optimum codes can be 100 times faster than before.
I will show that such codes exist today and both allow and require online visualization and interaction with the user.
Based on the example of particle-mesh methods I will present the most important techniques on how to write optimum codes.
Keywords: particlemesh; gpu; exaflop
Beteiligte Forschungsanlagen
- Strahlungsquelle ELBE DOI: 10.17815/jlsrf-2-58
Verknüpfte Publikationen
- DOI: 10.17815/jlsrf-2-58 is cited by this (Id 18695) publication
-
Eingeladener Vortrag (Konferenzbeitrag)
Seminar, Center of Excellence for Computational Fluid Dynamics, 12.04.2013, Warszawa, Polska
Permalink: https://www.hzdr.de/publications/Publ-18695