An OpenCL-based parallel CFD code for simulations on hybrid systems with massively-parallel accelerators

摘要

A parallel finite-volume CFD algorithm for modeling of incompressible flows on hybrid supercomputers is presented. It is based on\uda symmetry-preserving high-order numerical scheme for structured meshes. A multilevel approach that combines di erent parallel\udmodels is used for large-scale simulations on computing systems with massively-parallel accelerators. MPI is used on the first\udlevel within the distributed memory model to couple computing nodes of a supercomputer. On the second level OpenMP is used to\udengage multiple CPU cores of a computing node. The third level exploits the computing potential of massively-parallel accelerators\udsuch as GPU (Graphics Processing Units) of AMD and NVIDIA, or Intel Xeon Phi accelerators of the MIC (Many Integrated Core)\udarchitecture. The hardware independent OpenCL standard is used to compute on accelerators of di erent architectures within a\udgeneral model for a combination of a central processor and a math co-processor.