Chair for Practical Computer Science






SFB 393: Numerical Simulation on Massively Parallel Computers

B8: Parallelization of Irregular Numerical Algorithms

The class of irregular algorithms includes applications with sparse, grid-based, and temporally varying data structures as well as applications with irregular and run-time dependent computation and communication structures. The aim of this project is the parallelization of different algorithms particularly with regard to an efficient implementation for heterogeneous platforms, like clusters of SMPs (symmetric multiprocessors).

Current work:
  • Task Pool Teams for irregular algorithms, like the Hierarchical Radiosity Algorithm
  • Modular communication and data distribution support on software level for finite element methods (FEM) with adaptive mesh refinement (AMR)

Joint work with

Department of Computer Science, Chemnitz University of Technology
Prof. Dr. Gudula Rünger
Judith Hippold
Michael Schwind