高级检索
赵鹏, 沈弼龙, 王大伟, 熊志輝, 李思昆. 面向异构MPSoC的多媒体应用程序任务分配方法[J]. 计算机辅助设计与图形学学报, 2010, 22(10): 1671-1678.
引用本文: 赵鹏, 沈弼龙, 王大伟, 熊志輝, 李思昆. 面向异构MPSoC的多媒体应用程序任务分配方法[J]. 计算机辅助设计与图形学学报, 2010, 22(10): 1671-1678.
Zhao Peng, Shen Bilong, Wang Dawei, Xiong Zhihui, Li Sikun. Research on Task Allocation of Multimedia Applications onto Heterogeneous Multi-processor System-on-Chip[J]. Journal of Computer-Aided Design & Computer Graphics, 2010, 22(10): 1671-1678.
Citation: Zhao Peng, Shen Bilong, Wang Dawei, Xiong Zhihui, Li Sikun. Research on Task Allocation of Multimedia Applications onto Heterogeneous Multi-processor System-on-Chip[J]. Journal of Computer-Aided Design & Computer Graphics, 2010, 22(10): 1671-1678.

面向异构MPSoC的多媒体应用程序任务分配方法

Research on Task Allocation of Multimedia Applications onto Heterogeneous Multi-processor System-on-Chip

  • 摘要: 任务分配是多处理器SoC功能实现与性能优化的重要步骤,严重影响着多处理器SoC系统的处理性能与效率.文中针对多媒体应用程序向异构多处理器SoC的任务分配问题,提出图结点多着色模型来描述任务分配问题,并使用进化蚁群算法进行任务分配.在任务分配的过程中,首先对多媒体应用进行预处理,包括应用特征分析、并行任务划分与功能模型生成;然后启动进化蚁群算法进行分配空间探索,直至找到满足条件的高质量任务分配方案.实验结果表明,相对于采用基本蚁群算法与遗传算法的任务分配方法,文中方法可以获得高质量的分配方案,并较大幅度地加快了任务分配空间探索的收敛速度.

     

    Abstract: Task allocation is an important step in the process of multi-processor system-on-chip (SoC) implementation and optimization.And task allocation greatly influences SoC performance and efficiency.This paper proposes a method for automatically allocating multimedia processing tasks onto heterogeneous multi-processor SoC.The task allocation problem is defined using graph node coloring models.And the proposed evolutional ant colony algorithm is applied to find the high-quality allocation scheme.In the process of task allocation, the multimedia processing programs is firstly preprocessed, including program feature analysis, parallel task partition and function model generation.And then, the evolutional ant colony algorithm is used to find the high-quality allocation scheme iteratively.The experimental results show that the proposed method can find better allocation scheme within less time.

     

/

返回文章
返回