Performance Evaluation of Image Processing Algorithm on GPU

Abstract

CUDA stands for Compute Unified Device Architecture is a parallel comput- ing architecture developed by NVIDIA. CUDA is the computing engine in NVIDIA graphics processing units or GPUs that is accessible to software developers through industry standard programming languages. GPUs are emerging as platform of choice for high performance parallel computing. GPUs are good at data intensive parallel processing with availability of software development platforms such as CUDA (de- veloped by NVIDIA for its GeForce series GPUs). There is thrust in using GPUs for main stream high performance computing. Basic goal of CUDA is to help pro- grammers focus on the task of parallelization of the algorithms rather than spending time on their implementation. The parallel versions of 1D, 2D Complex to Com- plex Fourier transformation & Real to Complex Fourier transformation have been developed on GeForce GT 130 on Mac OS. The results are presented along with methodology for parallelization of code for NVIDIA GPU. It is well suited to address problems that can be expressed as data parallel computations. The same program is executed on many FFT elements using kernel functions & streams. In this thesis it is used for image processing application like sobel edge detection filter to achieve a faster implementation of it using CUDA and OpenGL.