【摘要】 基于高效能大任务和多任务计算需求以及芯片技术的高速发展,多核处理器进入片上众核处理器时代,但如何将这种硬件能力转变成计算性能的提升,充分利用MPSo C的并行计算能力是该领域亟待解决的问题.文中研究设计了面向片上众核处理器的物理核到虚拟核分配算法(VPD),该算法通过识别任务执行时的阶段状态特征,动态重组物理核并配置成与任务计算需求相适应的虚拟核,在满足任务的计算资源需求的条件下,实现虚拟核最优分配.最后通过实验将VPD算法与不同的多核/众核处理器环境下的主流调度算法进行比较,结果表明:该算法可适用于多种环境,且效能比其他同类算法平均高5%~10%.更多还原

【Abstract】 Based on the high efficiency large task and multi-task computing needs and speedy development of the chip technology,multi-core processors enter the era of Many-core Processor System-on-Chip,how to turn this hardware capability into a computational performance improvement and make full use of the parallel computing power of MPSo C is an urgent problem in this field. The purpose of this paper is to study the design of a physical core to virtual core allocation algorithm( VPD) for Many-core Processor System-on-Chip,the algorithm dynamically reconfigures the physical cores by recognizing the phase state characteristics of task execution,and configures the virtual cores according to the task computing requirements. Under the condition of meeting the computing resource requirements of task,the algorithm achieves optimal allocation of virtual cores. Finally,in the experiment,VPD algorithm is used to compare with several mainstream scheduling algorithms in different multi-core/many-core processors environments. Experimental results show that: the algorithm can be applied to a variety of environments,and the efficiency is higher than that of other similar algorithms 5% ~ 10%.更多还原