一种新的异构多核平台下多类型DAG调度方法  

Novel scheduling method for multi-type DAG on heterogeneous multi-core platforms

作　　者：左俊杰 肖锋[1] 黄姝娟[1] 沈超 郝鹏涛 陈磊 Zuo Junjie;Xiao Feng;Huang Shujuan;Shen Chao;Hao Pengtao;Chen Lei(School of Computer Science&Engineering,Xi’an Technological University,Xi’an 710021,China)

机构地区：[1]西安工业大学计算机科学与工程学院,西安710021

出　　处：《计算机应用研究》2025年第2期514-518,共5页Application Research of Computers

基　　金：国家自然基金面上项目(62171361);陕西省科技厅重点研发计划资助项目(2023-YBGY-027);陕西省教育厅专项科研计划资助项目(22JK0412)。

摘　　要：异构多核处理器在异构环境中受限于处理器种类,只能在特定处理器上执行。现有调度方法通常使用多类型DAG(directed acyclic graph)任务模型进行模拟,但调度方法往往忽略不同核上的通信开销,或未考虑处理器与节点的对应关系,导致调度时间开销较大,处理器资源未充分利用,任务效率低。针对上述问题,提出了PNIF(processor-node impact factor)算法。该算法引入了两个对节点优先级具有重大影响的比例因子,将它们加入到节点优先级的计算中从而确定任务执行顺序。实验结果表明,PNIF比PEFT、HEFT、CPOP在调度长度上分别平均提升5.902%、19.402%、25.831%,有效缩短了整体调度长度,提升了处理器资源利用率。In a heterogeneous environment,heterogeneous multi-core processors are limited to processor types and can only exe-cute on specific processors.The existing scheduling methods usually use the DAG task model to simulate,but they often ignore the communication overhead on different cores,or do not consider the corresponding relationship between processors and nodes,which leads to large scheduling time overhead,underutilization of processor resources,and low task efficiency.This paper proposed PNIF(processor-node impact factor)algorithm to solve the above problems.The algorithm introduced two scale factors which had a significant impact on the priority of the node,and added them to the calculation of the node priority to determine the task execution order.Experimental results show that compared with PEFT,HEFT and CPOP,PNIF increases the schedule length by 5.902%,19.402%and 25.831%on average respectively,which effectively shortens the overall schedule length and improves the processor resource utilization.

关 键 词：异构多核处理器 多类型DAG任务 任务调度 影响因子 PNIF算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]