异构多处理平台并行实时编码算法研究  

Realtime Parallel Algorithm Study on Heterogeneous Multiprocessing Platform

作　　者：袁三男 王孟彬 陶倩昀 张艳秋 YUAN Sannan;WANG Mengbin;TAO Qianyun;ZHANG Yanqiu(School of Electronic and Information Engineering,Shanghai University of Electric Power,Shanghai200090,China)

机构地区：[1]上海电力大学电子与信息工程学院,上海200090

出　　处：《上海电力大学学报》2022年第2期158-162,共5页Journal of Shanghai University of Electric Power

摘　　要：针对以CPU为处理平台的H.265/HEVC串并行编码效率较低的问题,设计了一种基于异构多核CPU+GPU处理平台的并行实时编码算法。按照CPU和GPU互不相同的硬件特性分配任务,降低了时间复杂度,同时使CPU和GPU的协作能力获得了提升,计算资源得到了更加合理的利用。视频编码并行化设计提高了编码效率,高清视频的编码速度最高可达26.31帧/s,实现了高清视频的实时编码。Aiming at the low efficiency of H.265/HEVC serial parallel coding based on CPU,a parallel real-time coding algorithm based on heterogeneous multi-core CPU+GPU processing platform is designed.According to the different hardware characteristics of CPU and GPU,tasks are allocated,which reduces the time complexity,improves the cooperation ability of GPU and CPU,and makes more reasonable use of computing resources.The parallel design of video coding improves the coding efficiency and makes the coding speed of high-definition video up to 26.31 fps,realizing the real-time coding of high-definition video.

关 键 词：实时编码 异构多处理平台 并行化设计 

分 类 号：TN949.6[电子电信—信号与信息处理]