Finite element explicit parallel solving and simulating method based on graphic processing unit (GPU)

Abstract

The invention discloses a finite element explicit parallel solving and simulating method based on a graphic processing unit (GPU). The method comprises the following practicing steps: 1) data is read into a host memory; nodes and units in finite element calculating model data are recognized through a host central processing unit (CPU); the ownership between the nodes and the units is searched and stored, and the finite element calculating model data is read into a GPU global memory; 2) a GPU executing parameter is set; 3) data stored in the GPU global memory is read through the GPU, the node acting force of each node is calculated through a multithreading method according to the GPU executing parameter; and 4) the calculated result is selectively output; CPU determines to quit or return to execute step 3) according to the convergence and judgment conditions. The finite element explicit parallel solving and simulating method is quick in operation speed, low in calculating cost, low in hardware cost and simple and convenient to use; particularly, the method is applicable to the dynamic nonlinear calculation of physical fields including sheet forming simulating optimization, automobile crash simulation and the like..

Description

technical field [0001] The invention relates to the field of finite element simulation in computer aided engineering, in particular to a GPU-based finite element explicit parallel solution simulation method. Background technique [0002] The explicit finite element calculation method does not need to solve any equations, and has good stability and convergence. It has been widely used in the simulation calculation of strong nonlinear problems such as automobile collision and sheet metal forming. With the increasing complexity of modern engineering problems, the scale of finite element calculation models is getting larger and larger. For explicit finite element algorithms with stable conditions, the calculation time step is generally at the microsecond level, while the actual simulation time is generally at the second level. means a huge amount of calculation. [0003] With the development of computing technology, parallel computing method is an effective means to solve the p...

AI Technical Summary

Problems solved by technology

The parallel finite element calculation method in the prior art mainly adopts parallel processing technologies such as distributed computing, parallel computer or multi-threading on the technical level, and the computing hardware used is mainly a parallel computing platform with CPU as the computing core. While the speedup ratio is good, there are also the following disadvantages: On the one hand, for large-scale finite element models, the large scale of units and nodes leads to an increase in communication consumption between distributed computers or between multi-threads, making it difficult to improve the speedup ratio; On the other hand, parallel machines are expensive, relatively complicated to use and maintain, and require high algorithm rewriting and programming skills for researchers, making it difficult to popularize

In addition, the current general-purpose computing platform based on GPU and CUDA has low hardware cost, easy software implementation, and obvious improvement in computing efficiency. It has become a new research hotspot at home and abroad. At present, in the field of mechanics, explicit finite element A general-purpose computing platform of GPU and CUDA to achieve parallel solution simulation

Images