Optimization and parallelization of the DSMC method on unstructured grids

Matthias Laux, Matthias Laux · 32nd Thermophysics Conference · 1997

This paper describes the optimization and parallelization of the LasVegas package initially developed at the Space Systems Institute of the University of Stuttgart. Optimization aspects for the DSMC method on unstructured grids in 2 spatial dimensions are described with respect to superscalar cache-based RISC architectures and the exploitation of FORTRAN90 features. The parallelization approach described here is based on domain decomposition and the message passing programming paradigm using the MPI standard.

Read the paper · More papers on PaperTik