Abstract:
We analyze the performance of some computational models of filtration gas combustion on multi-core computers. The analysis is limited to models based on explicit difference schemes. In particular, an explicit two-level parallel algorithm using an adaptive mesh is constructed The two ways of shared memory parallelization are applied: direct application of OpenMP directives and a special data distribution between the threads. It is shown numerically that the latter method provides a significant performance advantage.
Keywords:combustion wave, explicit scheme, adaptive mesh, parallelization on shared memory.