多核系统中一种改进的任务分配方案研究  

Research on improved task allocation scheme in many-core systems

作　　者：罗群[1] 

机构地区：[1]重庆科创职业学院信息工程学院,重庆402160

出　　处：《计算机应用研究》2015年第6期1732-1736,共5页Application Research of Computers

摘　　要：多核系统工作负载的动态性和不可预知性往往会导致系统功耗高、延时长,运行期间敏捷的任务分配方法能有效解决上述问题。为此,针对多核系统中的任务分配问题提出一种近似模型,以估计任意给定节点周围的可用节点数量;然后,采用爬坡搜索启发式策略(SHi C),在所有可用节点中迅速搜索出最优首个节点;最后,利用Co NA算法实现任务的高效分配。在不同网络规模和网络参数设置下展开仿真,实验结果表明,SHi C可实现显著的性能提升,与当前最新研究相比,改进了网络延时和功耗。Dynamic and unpredictable of multi-core system＇ s workload often lead to high power consumption and long delay. Agile task allocation during runtime can effectively solve this problem. Aiming at the task allocation in multi-core system , this paper proposed an approximate model to estimate the number of the available nodes around any given node. Then, this paper adapted a hill climbing search heuristic to find the optimum first node rapidly among all the available nodes. Finally, the task allocation was efficiently carried out through the CoNA algorithm. Simulation results over different network sizes and different network parameters show significant improvement influenced by SHiC, which results in better network latency and power dissi- pation, compared to state-of-the-art works.

关 键 词：多核系统 映射 爬坡搜索 功耗 延时 

分 类 号：TP302.7[自动化与计算机技术—计算机系统结构]