Research on Viterbi Algorithm Based on OpenCL Architecture

• Main Authors: Kuan-Hsien Lee, 黎冠賢
• Other Authors: Wei-Tyng Hong
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/7wds4t

碩士 === 元智大學 === 通訊工程學系 === 105 === 　　For the past few years, the capability of the computer graphic cards has been increasing year by year, they become capable of performing wide range of applications. In the past, we usually used CUDA (Compute Unified Device Architecture) syntax but Now there are numerous applications like OpenCL (Open Computing Language). 　　The main stumbling block for this increase in graphic card capabilities is the amount of the computer resources available. The general processing power of the CPU (Central Processing Unit) is somehow limited, but with the introduction of GPU (Graphics Processing Unit) base on parallel processing, has improved the speed of operation which enhanced the voice recognition applications greatly. 　　This paper will introduce OpenCL in detail and apply it to the Viterbi algorithm for acceleration. The purpose is to operate more rapidly on voice recognition in the future. 　　In this urban society, everything is required to be rapid and efficient, therefore making a good use of the application capabilities of modern graphic cards is of huge significant. Keywords:OpenCL, Viterbi.