{"title":"A Cache Hardware Design for H.264 Encoder","authors":"S. Zuo, Mingjiang Wang, Liyi Xiao","doi":"10.1109/IMCCC.2012.221","DOIUrl":null,"url":null,"abstract":"On the oretical basis of the inter-frame motion compensation of H.264 encoder, this paper proposed a new cache which adopted the LFU(lease frequently used) replace mechanism and a new prefetching method of reference images which have been reconstructed in H.264 encoder. We first gave its top-level framework definition, and then implemented the design using Verilog HDL. Based on the RTL coding, the design is synthesized with TSMC 90nm technology library. The result of the related experiments shows that the proposed cache can effectively save 76%~86% of the data memory bandwidth by use of the temporal correlation of data. Compared with the traditional method, the new cache architecture shows much better performance on reducing the bandwidth of accessing main storage, and presents much lower power consumption, higher hit ratio and faster access speed.","PeriodicalId":394548,"journal":{"name":"2012 Second International Conference on Instrumentation, Measurement, Computer, Communication and Control","volume":"9 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2012-12-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"3","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2012 Second International Conference on Instrumentation, Measurement, Computer, Communication and Control","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/IMCCC.2012.221","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 3
Abstract
On the oretical basis of the inter-frame motion compensation of H.264 encoder, this paper proposed a new cache which adopted the LFU(lease frequently used) replace mechanism and a new prefetching method of reference images which have been reconstructed in H.264 encoder. We first gave its top-level framework definition, and then implemented the design using Verilog HDL. Based on the RTL coding, the design is synthesized with TSMC 90nm technology library. The result of the related experiments shows that the proposed cache can effectively save 76%~86% of the data memory bandwidth by use of the temporal correlation of data. Compared with the traditional method, the new cache architecture shows much better performance on reducing the bandwidth of accessing main storage, and presents much lower power consumption, higher hit ratio and faster access speed.
在H.264编码器帧间运动补偿的理论基础上,提出了一种采用LFU(lease frequency - used)替换机制的缓存,并对H.264编码器重构后的参考图像进行预取的新方法。首先给出了其顶层框架定义,然后使用Verilog HDL实现了设计。在RTL编码的基础上,采用台积电90nm工艺库进行了综合设计。实验结果表明,该缓存利用数据的时间相关性,可有效节省数据存储带宽的76%~86%。与传统方法相比,该方法在减少访问主存的带宽方面表现出更好的性能,并且具有更低的功耗、更高的命中率和更快的访问速度。