基于算术编码的异构双核SoC代码压缩方法  

Code Compression Approach for Heterogeneous Dual-Core SOC Based on Arithmetic Coding

作　　者：阳晔[1] 葛海通 严晓浪[1] 杨军[1] 

机构地区：[1]浙江大学超大规模集成电路设计研究所,杭州310027 [2]杭州中天微系统公司,杭州310012

出　　处：《微电子学》2007年第6期857-861,共5页Microelectronics

基　　金：国家高技术研究发展(863)计划专项经费资助项目(2005AA1Z1271)

摘　　要：介绍了一种基于二进制算术编码和马尔可夫模型的代码压缩算法,并以此为基础,面向自主研发的异构双核SoC—GEM-SOC,提出了一种软件实现的解码方法。该方法可以根据不同应用的代码特点,选择合适的马尔可夫模型参数,从而达到最佳的代码压缩率。一组媒体应用测试表明,该方法可以显著减小双核SoC中DSP处理器的代码尺寸。A code compression approach for heterogeneous dual-core SOC is presented, which is based on binary arithmetic coding in combination with semi-adaptive Markov model. A software-based decompression scheme is also proposed, which can easily achieve the best compression ratio by adjusting the parameter of Markov model according to different applications. Experiments on self-developed heterogeneous dual-core SOC--GEM-SOC show this approach can effectively reduce the code size of DSP core.

关 键 词：算术编码 马尔可夫模型 代码压缩 异构双核 SOC 

分 类 号：TP332[自动化与计算机技术—计算机系统结构]