GPGPU implementation of FIC using texture memory

摘要

Fractal Image Coding (FIC) is a lossy compression technique with promising features, however it has been relegated due to its large coding time. The present article proposes an efficient parallel implementation of FIC algorithm on a general propose computing on graphic processing unit (GPGPU) system. Two parallel implementations using different memory spaces are compared: global memory and texture memory. The experiments demostrated that texture memory achived a better acceleration factor than global memory of about 350 times.