用超长指令实现DCT的新算法  

New DCT Computation Algorithm for VLIW Architecture

作　　者：李学明[1] 李继[1] 

机构地区：[1]北京邮电大学信息工程学院,北京100876

出　　处：《电子学报》2003年第7期1074-1077,共4页Acta Electronica Sinica

基　　金：国家自然科学基金 (No 60 1 72 0 50 )

摘　　要：本文介绍一种新的DCT计算方法 ,它以现有的DCT快速算法为基础 ,利用超长指令的并行特征来提高DCT计算的性能 .仿真结果表明 :该方法的运算速度比普通的DCT计算方法提高 73 % ,即便同快速算法相比 ,也可以提高 2A new DCT computation algorithm is presented, which is based on existing fast DCT algorithm and fully utilizes the parallelism of very long instruction word (VLIW) architecture to achieve high speed performance. Simulation results show that this method can reduce more than 73% and 24% computing cycles, compared with normal and fast DCT computation algorithm respectively.

关 键 词：离散余弦变换 超长指令字 并行算法 

分 类 号：TN955.2[电子电信—信号与信息处理]