1Khimich, OM, 1Mova, VI, 1Nikolaichuk, ОO, 1Popov, OV, 1Chistjakova, TV, 1Tulchinsky, VG
1Glushkov Institute of Cybernetics, NAS of Ukraine
Sci. innov. 2018, 14(6):61-72
Section: Research and Engineering Innovative Projects of the National Academy of Sciences of Ukraine
Language: English

Introduction. Mathematical modeling with large volumes of data is an actual innovation problem in various spheres of human activity. For their effective computer research, it is necessary to use powerful computers and high-performance software.
Problem Statement. Models of processes studied on modern computers have approximate data, their mathematical properties are a priori unknown. However, the existing software does not take this into consideration. Modern parallel computers require high costs for production and use.
Purpose. To develop an intelligent personal computer based on Intel Xeon Phi processors of new generation and intelligent software for automatic research and solution of the main classes of problems of computational mathematics with approximate data.
Materials and Methods. The concept and methods for intellectualization of parallel computers of the Inparcom family, which are developed at the Glushkov Institute of Cybernetics of the NAS of Ukraine in cooperation with Electronmash. 
Results. Intelligent Parallel Computer Inparcom_xp with an Intel Xeon Phi 7210 processor, which makes computations (up to 3.5 TFlops) in the single-node format (test model, prototype). Intelligent software for automatic research and solution of problems in computational mathematics.
Conclusions. The Inparcom_xp guarantees a high reliability of computer solutions of problems, frees users from creating parallel algorithms and programs. The computer is made mainly for individual use, thereby raises the resource of personal computing for R&D calculations.

Keywords: approximate data, computational mathematics, Intel Xeon Phi processor, mathematical modeling, parallel computer