This title appears in the Scientific Report : 2013 

Parallel Brownian Dynamics Simulation with MPI, OpenMP and UPC
Teijeiro, Carlos (Corresponding author)
Sutmann, Godehard / Taboada, Guillermo L. / Tourino, Juan
Jülich Supercomputing Center; JSC
Hybrid Particle-Continuum Methods in Computational Materials Physics
Jülich John von Neumann Institute for Computing (NIC) 2013
25 - 40
978-3-89336-849-5
Hybrid Particle-Continuum Methods in Computational Materials Physics, Jülich (Germany), 2013-03-04 - 2013-03-07
Contribution to a book
Contribution to a conference proceedings
Computational Science and Mathematical Methods
NIC Series 46
This work presents the design and implementation of a parallel simulation code for the Brownian motion of particles in a fluid. Three different parallelization approaches have been followed: (1) traditional distributed memory message-passing programming with MPI, (2) a directivebased approach on shared memory with OpenMP, and (3) the Partitioned Global Address Space (PGAS) programming model, oriented towards hybrid shared/distributed memory systems, with the Unified Parallel C (UPC) language. According to the selected environment, different domain decompositions and work distributions are studied in terms of efficiency and programmability in order to select the most suitable strategy. Performance results on different testbeds and using a large number of threads are presented in order to assess the performance and scalability of the parallel solutions.