Skip to Main content Skip to Navigation
Master Thesis

Étude et tests de CUDA™ (NVIDIA) appliqués à la mise en œuvre d’algorithmes de télécommunications

Abstract : This document presents the study of the compute resources of a video card from the most recent ones in order to observe to which degree it is able to replace the CPU for parallel computing. This document also presents an example of implementation of GPU computing for radar signal processing. In order to achieve this goal, we carried out the complete study of three APIs dedicated to computing on NVIDIA video cards : CUDA, CUBLAS and CUFFT. For this study, we used a GTX470 video card made by NVIDIA and featuring 448 streaming processors for parallel computing. The MATLAB software has been used in order to check the accuracy of the GPU computing. A MEX function has been used in order for MATLAB and the CUDA program to be able to exchange data. This study has been carried out using the MATLAB 7.1 and Visual C++ (Visual Studio 2005) softwares as well as the CUDA dedicated tools which are the Compute Visual Profiler and the Occupancy Calculator.
Document type :
Master Thesis
Complete list of metadata

Cited literature [10 references]  Display  Hide  Download

https://dumas.ccsd.cnrs.fr/dumas-01154150
Contributor : Cnam - Service Commun de la Documentation - Bibliothèque Centrale <>
Submitted on : Thursday, May 21, 2015 - 11:23:05 AM
Last modification on : Saturday, February 8, 2020 - 12:10:04 PM
Long-term archiving on: : Tuesday, September 15, 2015 - 6:37:22 AM

Identifiers

  • HAL Id : dumas-01154150, version 1

Citation

Jean-Charles Paoletti. Étude et tests de CUDA™ (NVIDIA) appliqués à la mise en œuvre d’algorithmes de télécommunications. Automatique / Robotique. 2013. ⟨dumas-01154150⟩

Share

Metrics

Record views

256

Files downloads

1267