Performance Evaluation of Fast Fourier Transform on CPU and GPU Environment

• Main Authors: LIU, CHUN-LIANG, 劉俊良
• Other Authors: Chu-Hsing Lin
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/10905196761939018084

碩士 === 東海大學 === 資訊工程學系 === 102 === Presently in order to strengthen varieties of 3D effects, computers are usually equipped with video adapters to speed up the processing of images and screen output. Compared to the Central Processing Unit (CPU for short) of serialization design, the Graphics Processing Unit (GPU for short) is specialized in graphic processing with a powerful parallel computing capability. Besides, Compute Unified Device Architecture (CUDA) developed by NVIDIA Corporation provides a parallel computing platform and software development model to enhance the computing power of the GPU. This thesis compares performance of GPU and CPU when performing the Fourier transform that performs a mathematic transform to transform data from the time domain into the frequency domain. We performed Fast Fourier Transforms for digital audio data and image in different file formats and file sizes by using GPU parallel computing to accelerate the processing time. Finally, we also compared the time efficiency of GPU and CPU in performing Fast Fourier Transforms for digital audios.