Abstract: OpenFOAM is a scientific library popularly used for solving and simulating Computational Fluid Dynamics (CFD) problem based on the finite volume method. A most recent hardware Intel Xeon Phi coprocessor introduced by Intel which is based on many integrated core (MIC) architecture. The MIC coprocessor has numerous cores than the traditional Intel Xeon processor but each core works at a lesser frequency and is less powerful compared to the Xeon counterpart. In this paper, we studied different hybrid and heterogeneous platform and discussed various methods for achieving good scalability across nodes on Intel MIC. We have discussed a few approaches for evaluating and optimizing the performance of message passing interface (MPI) to reduce the execution time of OpenFOAM solver on Intel MIC architecture.
Keywords: Intel MIC, Linear system solver, OpenFOAM, Optimization.