Hardware-oriented numerics and concepts for PDE software

Authors:
S. Turek;Chr. Becker;S. Kilian
Affiliations:
Institute for Applied Mathematics, LS III, University of Dortmund, Dortmund, Germany;Institute for Applied Mathematics, LS III, University of Dortmund, Vogelpothsweg, Dortmund, Germany;Institute for Applied Mathematics, LS III, University of Dortmund, Vogelpothsweg, Dortmund, Germany
Venue:
Future Generation Computer Systems
Year:
2006

Citing 5
Cited 4

A second-order accurate pressure correction scheme for viscous incompressible flow

SIAM Journal on Scientific and Statistical Computing
Performance rating via the FEAST indices

Computing
Parallel multigrid in an adaptive PDE solver based on hashing and space-filling curves

Parallel Computing - Special issue on parallelization techniques for numerical modelling
Sources of Parallel Inefficiency for Incompressible CFD Simulations (Research Note)

Euro-Par '02 Proceedings of the 8th International Euro-Par Conference on Parallel Processing
Consequences of Modern Hardware Design for Numerical Simulations and Their Realization in FEAST

Euro-Par '99 Proceedings of the 5th International Euro-Par Conference on Parallel Processing

Exploring weak scalability for FEM calculations on a GPU-enhanced cluster

Parallel Computing
Using GPUs to improve multigrid solver performance on a cluster

International Journal of Computational Science and Engineering
A precompiler to reduce the memory footprint of multiscale PDE solvers in C++

Future Generation Computer Systems
Co-processor acceleration of an unmodified parallel solid mechanics code with FEASTGPU

International Journal of Computational Science and Engineering

Quantified Score

Hi-index	0.00

Visualization

Abstract

Processor technology is still dramatically advancing and promises further enormous improvements in processing data for the next decade. On the other hand, much lower advances in moving data are expected such that the efficiency of many numerical software tools for partial differential equations (PDEs) are restricted by the cost for memory access. We demonstrate how data locality and pipelining can achieve a significant percentage of the available huge computing power, and we explain the influence of processor technology on recent and future numerical PDE simulation tools. Exemplarily, we describe hardware-oriented concepts for adaptive error control, multigrid/domain decomposition schemes and incompressible flow solvers and discuss their numerical and computational characteristics.