Rüde U (2018)
Publication Type: Conference contribution, Abstract of lecture
Publication year: 2018
Event location: Linz, Austria
Exploiting future extreme scale computers to their full capability will require innovation on many levels: New algorithmic paradigms must address unprecedented levels of concurrency in heterogenous systems. In this talk will report on our work towards extreme scale predictive simulations for physics and engineering problems. Our focus will be on coupled multifield problems as they arise in geophysics or in the direct numerical simulation of complex flows. When we model the whole volume of the planet with a finite element mesh of 1km resolution, this results in a linear system with more than a trillion (10^12) degrees of freedom. Similarly large models can be used with the lattice Boltzmann method for the direct simulation of particulate suspensions for environmental or biomedical flows. Based on these examples at the current limits of what can be simulated, we will discuss some challenges on the road to predictive science.
APA:
Rüde, U. (2018, February). Festvortrag: Supercomputing: Expanding the limits of predictability. Paper presentation at Inbetriebnahme des Supercomputers MACH-2, JKU Linz, Austria, Linz, Austria.
MLA:
Rüde, Ulrich. "Festvortrag: Supercomputing: Expanding the limits of predictability." Presented at Inbetriebnahme des Supercomputers MACH-2, JKU Linz, Austria, Linz, Austria 2018.
BibTeX: Download