CORC  > 清华大学
多视点视频编码中的SIMD优化研究
李彦洁 ; 庞一 ; 孙立峰 ; Li Yan jie ; Pang Yi ; Sun Li
2010-07-15 ; 2010-07-15
会议名称第四届和谐人机环境联合学术会议 ; CNKI
关键词SIMD 多视点视频编码 CELL DCT 预测模式 SIMD Multi-View Video Coding(MVC) CELL DCT Prediction modes TN919.81
其他题名MULTI-VIEW VIDEO CODING OPTIMIZATION EXPLOITING SIMD INSTRUCTIONS
中文摘要多视点视频是当前研究的热门领域,应用前景比较广泛。本文中主要讨论基于IBM的CELL多核处理机的多视点视频编码(Multi-view Video Coding,简称MVC)中CELL支持的SIMD指令的应用。本文总结了一些使用SIMD指令进行优化的思想,分析使用SIMD指令的预期效果。依据这些SIMD优化思想,本文提出一个SIMD优化算法,来实际说明SIMD优化的应用。并且通过列举实验数据说明在DCT和帧间预测模式两部分中使用SIMD优化之后,此部分的速度为未使用SIMD的情况下的大约2.28倍。从而证明SIMD指令的应用潜能。同时本文虽然基于MVC,但是总结的SIMD使用思想,对CELL上的SIMD优化具有普遍意义。; In this paper, Multi-View Video Coding scheme was speeded up by using CELL’s SIMD instructions. This paper summarizes some ideas of using CELL’s SIMD instructions, and gives an optimization method that has been practically experimented on the Cell and received some good results. This method may be could be utilized some other place because DCT is a common transfer that needed not only in MVC scheme in this paper. This paper would show the MVC’s possibility of speeding up by SIMD instructions. This paper gives some results of the speeding up by SIMD instructions that the speed after using SIMD instruction is about 2.28 times the speed before utilizing SIMD instructions. Although only DCT part has done experiment, it has proved that MVC has a good property to have data-level parallel and that SIMD is a good way to do optimization. As the developing of hardware, it is more and more practical to utilize SIMD instructions to speed up progress. And the possibility of parallel should be an important requirement of new coding and encoding standard.
语种中文 ; 中文
内容类型会议论文
源URL[http://hdl.handle.net/123456789/69975]  
专题清华大学
推荐引用方式
GB/T 7714
李彦洁,庞一,孙立峰,等. 多视点视频编码中的SIMD优化研究[C]. 见:第四届和谐人机环境联合学术会议, CNKI.
个性服务
查看访问统计
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。


©版权所有 ©2017 CSpace - Powered by CSpace