Two-stage partitioned two-time polycondensation parallel computing system development method and parallel computing system

Abstract

The invention provides a two-stage partitioned two-time polycondensation parallel computing system development method and a corresponding parallel computing system. The two-stage partitioned two-time polycondensation parallel computing system development method comprises the steps of 1 producing all program source codes required by finite element analysis parallel computing through a finite element program automatic generation system according to a mathematical equation expression for solving problems, 2 removing solution linear equation systems in the program source codes and conducting parallelized modification on remaining modules by means of a solution linear equation, 3 developing parallel solution linear equation systems based on two-stage partitioning and two-time polycondensation strategies and conducting seamless integration on the parallel solution linear equation systems and the modules subjected to parallelized modification, and 4 compiling and debugging the integrated system and generating a finite element analysis parallel computing executable program. By means of the two-stage partitioned two-time polycondensation parallel computing system development method and the corresponding parallel computing system, solutions can be provided for the finite element analysis problems in all disciplines and fields, and the quality and efficiency of finite element analysis parallel computing software development can be improved.

Description

technical field [0001] The invention relates to a design method in the field of supercomputer application technology, in particular to a development method for a parallel computing system of two-stage partitioning and two polycondensation. Background technique [0002] Although my country has made many important breakthroughs in the development of supercomputers in recent years, it lags far behind the development of hardware in the development and application of high-performance computing software. In order to meet the needs of domestic scientific research and engineering analysis, our country has to pay expensive usage fees for foreign large-scale commercial software every year. In the field of finite element analysis, well-known foreign large-scale commercial software such as Ansys, Abaqus, Ls-Dyna and Adina have released parallel computing versions. However, due to various reasons, they impose a series of restrictions on the maximum number of parallel solving cores that ...

AI Technical Summary

Problems solved by technology

First of all, the architecture of SiPESC.FEMS is too large, so it is not easy to figure out the functions of each part of the program and the calling relationship between subroutines in a short time, and it is difficult to modify the program or add new functions on this basis bigger

Secondly, the current functions of SiPESC.FEMS are still very limited, so the parallel computing software developed on the basis of it is only suitable for solving specific fields and specific types of finite element analysis problems

Images