摘要
在多核处理器蓬勃发展的今天,温度过高成为制约其性能和稳定的关键因素。该文在单核处理器上,以热传递理论为基础,通过温度与时间的一个简明等式,详细分析了任务组的各种排列方式对单核处理器的峰值温度可能造成的影响,并提出了简单易行免于复杂计算的调度方法;然后将该方法拓展到多核处理器环境,通过合理分配、核上调度和核间迁移,达到了降低各核峰值温度的目的。最后通过实验表明,该方法是有效的。
Multi-core Processors are flourishing today, but high temperature has been the key factor that restrict the performance and stability processor. In this paper, we firstly analyze the effect of any permutation of the tasks to the peak temperature of one-core processor based on the theory of heat transfer and a simple equation of temperate and time, and then we propose a simple scheduling method to be free of complex computation.. And then we expand this method to multi-core environment and reduce the peak temperature of every core by distributing, scheduling and migrating tasks. At last, experiment results prove the method to be effective.
作者
宋鹏飞
曾国荪
SONG Peng-fei, ZENG Guo-sun (Department of Computer Science and Technology, Tongji University, Shanghai 201804, China)
出处
《电脑知识与技术(过刊)》
2010年第17期4799-4802,4808,共5页
Computer Knowledge and Technology
基金
863项目(2007AA01Z425,2009AA012201)
973计划课题(2007CB316502)
国家自然基金项目(90718015)
NSFC-微软亚洲研究院联合资助项目(60970155)
教育部博士点基金项目(20090072110035)
上海市优秀学科带头人计划项目(10XD1404400)
高效能服务器和存储技术国家重点实验室开放基金项目(2009HSSA06)
关键词
温度感知
热传递
多核
任务调度
temperature-aware
heat transfer
multi-core
scheduling