A comparative analysis of the parallel cluster multiple labeling technique on various sections of the MVS-10P OP supercomputer

The paper provides a comparative analysis of the Parallel Cluster Multiple Labeling Technique on five different sections of the MVS-10P OP supercomputer (taking into account the addition of a new section in 2021 and modernization of existing ones) installed at the JSCC RAS. At the JSCC RAS, the Parallel Cluster Multiple Labeling Technique is used to study the processes of epidemic spread. At the same time, it is a versatile tool that can be used in any field as a tool for differentiating large lattice clusters receiving data as input in an application-independent format. There are known developments using this algorithm to study the processes of water flow through porous materials, the behavior of oil reservoirs, and the spread of forest fires. The supercomputer simulation experiment involved the improved version of the technique for multiple labeling of Hoshen-Kopelman percolation clusters associated with the labels linking mech-anism improved for using on a multiprocessor system. The paper provides a comparative analysis of the execution time of the algorithm for multiple marking of Hoshen-Kopelman percolation clusters at full load of computing nodes and different values of input parameters on five partitions (Broadwell, Cascadelake, Skylake, Optan, KNL) of the MVS-10P OP supercomputer installed at the Interdepartmental Supercomputer Center of the Russian Academy of Sciences.