Optimization of a finite element code implemented in MATLAB. On the use of GPUs for High Performance Computing
[EN] The Department of Mechanical and Materials Engineering has developed a 2D Finite Element code based on geometry independent Cartesian grids (cgFEM) capable of solving shape optimization problems as well as making patient-specific analyses using medical images. A similar code in 3D (FEAVox) is c...
| Autor: | |
|---|---|
| Formato: | tesis de maestría |
| Fecha de publicación: | 2014 |
| País: | España |
| Recursos: | Universitat Politècnica de València (UPV) |
| Repositorio: | RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia |
| Idioma: | inglés |
| OAI Identifier: | oai:riunet.upv.es:10251/53393 |
| Acesso em linha: | https://riunet.upv.es/handle/10251/53393 |
| Access Level: | acceso abierto |
| Palavra-chave: | Método de los elementos finitos Mallados cartesianos Estructura de datos jerárquica Computación en paralelo HPC GPU MATLAB Finite element method Cartesian grids Hierarchical data structure Parallel computing INGENIERIA MECANICA Máster Universitario en Ingeniería Mecánica y Materiales-Màster Universitari en Enginyeria Mecànica i Materials |
| Resumo: | [EN] The Department of Mechanical and Materials Engineering has developed a 2D Finite Element code based on geometry independent Cartesian grids (cgFEM) capable of solving shape optimization problems as well as making patient-specific analyses using medical images. A similar code in 3D (FEAVox) is currently under development. Both codes are implemented in MATLAB, a simple and intuitive programming language but with a higher computational cost than compiled languages such as C++ or FORTRAN. The objective of this Thesis is to develop programming procedures to improve the performance of the existing and the currently under development software. Among other optimization techniques this Thesis will focus on the use of Graphics Processing Units (GPU) for high performance computing. The use of these techniques has led to a software that, despite being implemented with MATLAB, improves the computational efficiency of commercial software which is developed using compiled programming languages. |
|---|