High Performance Computing: from sequencial programming in CPUs to parallel codes in GPUs

• Grant number: 11/21265-7
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: February 01, 2012
• End date: January 31, 2013
• Field of knowledge: Engineering - Mechanical Engineering - Mechanics of Solids
• Principal Investigator: Ronaldo Carrion
• Grantee: Guilherme Henrique Pereira Silva
• Host Institution: Escola de Engenharia de São Carlos (EESC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

High-Performance Computing (HPC) has become necessary in various academic and scientific fields and engineering, but the current processor limitation has brought several restrictions in solving problems with large amounts of calculations including floating points and arrays. A recent solution for this limitation is the use of General Purpose Graphics Processing Units (GPGPU) which have shown high efficiency in parallel processing. This undergraduate research project aims to develop sequential codes in C language, dealing with structural analysis and based on numerical methods consecrated as the Finite Element Method (FEM) and Boundary Element (BEM). Besides that, another point to be reached in this work is the migration of these same codes for parallel processing to run on graphics units, with CUDA" developer environment. At the end of this work, it is intended to have some codes involving those numerical methods mentioned previously, applied to structural analysis. It is noteworthy these codes are representative examples of knowledge acquisition to develop and run codes on graphics units.(AU)