| Biological system simulation is an increasingly popular field of study that provides biologists with the tools necessary to simulate biochemical systems in order to obtain quantitative models. The purpose of this thesis is to describe an accelerated version of GridCell, a stochastic biological system simulator. GridCell tracks each individual particle's location in the system, as well as the time evolution of the concentration of each species involved. It simulates molecular diffusion via Brownian movements, and particle interactions are dependent on their locations. We present here a parallel adaptation of the algorithm, implemented on a heterogeneous multicore processor, i.e. IBM Cell Broadband Engine (CBE). We introduce the CBE architecture and outline its advantages, as well as describe the original algorithm. Subsequently, we detail the parallel implementation and the algorithm modifications. Finally, we perform timing analysis to show that the parallel version provides improved performance over the original serial version. |
