Scaling of Supercomputer Calculations on Unstructured Surface Computational Meshes

When solving complex problems of numerical modeling, computational meshes, containing hundreds millions of cells are quite often. Modern tasks even cross the line of billion cells. Workstations are unable to cope with such volume of data and computation. To perform computations of this volume we need to use supercomputer clusters consisting of many computational nodes interconnected by a high-speed communication network. In this case, it is necessary to perform the decomposition of the computational mesh into separate domains in order to ensure its parallel processing on all nodes of the cluster. These domains are distributed among the computational nodes of the supercomputer and are processed independently of each other. To efficiently perform calculations and scale them to a large number of computational nodes, it is necessary to develop efficient algorithms for decomposition of computational meshes that generate many domains with imposed requirements. We consider an hierarchical decomposition algorithm with the choice of the optimal criterion for dividing mesh into domains. As such a mesh we study an unstructured surface mesh used to calculate the processes of interaction of a volumetric body with the environment. Using this decomposition algorithm, supercomputer calculations are performed on the computing resources of JSCC RAS in order to measure the practical indicators of scalability of highly loaded applications.