{"title":"H.264 decoder optimization exploiting SIMD instructions","authors":"Juyup Lee, Sungkun Moon, Wonyong Sung","doi":"10.1109/APCCAS.2004.1413088","DOIUrl":null,"url":null,"abstract":"In this paper, H.264/AVC, the newest video coding standard, baseline profile decoder was implemented exploiting Intel MMX instructions. Both control- and data-level parallel processing approaches'were applied to the kernels of the baseline subsystems for efficiently utilizing the SIMD (Single Instruction Multiple Data) instructions. The data-level parallel approach tries to process multiple pixels at a time to fully utilize SIMD instructions. The data-level approach shows a better performance even though loop unrolling is further applied to the conlrol-level approach. The resultant implementations are also compared with the Intel Performance Primitives.","PeriodicalId":426683,"journal":{"name":"The 2004 IEEE Asia-Pacific Conference on Circuits and Systems, 2004. Proceedings.","volume":"12 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2004-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"The 2004 IEEE Asia-Pacific Conference on Circuits and Systems, 2004. Proceedings.","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/APCCAS.2004.1413088","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1
Abstract
In this paper, H.264/AVC, the newest video coding standard, baseline profile decoder was implemented exploiting Intel MMX instructions. Both control- and data-level parallel processing approaches'were applied to the kernels of the baseline subsystems for efficiently utilizing the SIMD (Single Instruction Multiple Data) instructions. The data-level parallel approach tries to process multiple pixels at a time to fully utilize SIMD instructions. The data-level approach shows a better performance even though loop unrolling is further applied to the conlrol-level approach. The resultant implementations are also compared with the Intel Performance Primitives.