Parallel Performance Studies for COMSOL Multiphysics Using Scripting and Batch Processing

N. Petra[1], and M.K. Gobbert[1]

[1]Department of Mathematics and Statistics, University of Maryland, Baltimore County, Baltimore, Maryland, USA
发布日期 2009

The graphical user interface (GUI) of COMSOL Multiphysics offers an effective environment to get started solving problems. For reproducibility of the results, it is often desirable to explore the script-based modeling capabilities of COMSOL with MATLAB. There are also potential benefits of running COMSOL in parallel, specifically by running several computational threads in shared-memory parallelism mode. We use the scripting abilities of COMSOL with MATLAB to study the shared-memory parallel performance of COMSOL, that is, the solution time required by using multi-threading on one multiprocessor, multi-core computer with shared memory among the processors. The performance results show that using more than one thread saves time, but the speed up is not in proportion to the number of cores used.