Search

Due to the physical limitations on the speed of individual processors, we are currently experiencing a paradigm shift in computer architecture towards multi-processor, multi-core and multi-subcore systems. In order to take full advantage of these new architectures, some of the basic models, algorithms and data structures in Molecular Dynamics simulations need to be re-thought. In this paper, I present some simple, generic algorithmic considerations that can improve the performance of such simulations on the IBM Cell Broadband Engine Architecture, a multi-subcore processor. The accumulated effects of these improvements are on par with what can be achieved with hand-tuned architecture-specific optimizations in the innermost loops of the simulation. [ABSTRACT FROM AUTHOR]

The requirement of high performance and memory for computer simulations is still growing. Due to hardware constraints like power consumption, heat dissipation and other physical limitations the development trend in high performance computing (HPC) tends to multicore design patterns. As new computational platforms become increasingly more complicated and heterogeneous, there is the need for portable programming models that easily enable the exploitation of these architectures. Additionally, algorithms are needed that are able to match the platform specific requirements and exploit their potential power. This work focuses on the particle-based algorithm Multiparticle Collision Dynamics (MPC) for the calculation of hydrody-namic properties of fluid and flow phenomena. This algorithm has already been ported to Cell Broadband Engine (Cell/BE) by using the high-level programming model Cell Superscalar (CellSs). Performance results of the Cell/BE implementation and a recently developed OpenMP version are presented. Furthermore, the possibilities for porting this application also to GPUs with minor effort are pointed out and a strategy for hybrid implementations to use multiple nodes in a cluster are examined. [ABSTRACT FROM AUTHOR]