首页> 外文会议>International Conference for Convergence in Technology >Task Migration Algorithm to Reduce Temperature Imbalance Amongst Cores in Linux Based Multi-Core Processor Systems
【24h】

Task Migration Algorithm to Reduce Temperature Imbalance Amongst Cores in Linux Based Multi-Core Processor Systems

机译:任务迁移算法减少基于Linux多核处理器系统中的核心温度不平衡

获取原文

摘要

Multi-core architectures have recently emerged as the main design paradigm for current and future processors, however, the addition of multiple cores on the same chip and simultaneous execution of processes on these cores increases the temperature imbalance amongst cores and result in an increase in chip temperature, which in turn reduces the reliability of the processor. In this research project, the problem of temperature imbalance amongst cores in multi-core processor systems is addressed at the operating system level via task migration. The proposed task migration algorithm finds out the thermal contribution of individual task and performs the task migration based on this thermal contribution, in order to maintain temperature balance amongst cores. The proposed task migration algorithm is implemented in Linux operating system by modifying the source code written in C language. Different categories of processes (e.g. CPU bound, IO bound etc.) are executed from Phoronix Test Suite and Sysbench Benchmarks to test the performance of the proposed algorithm. Experimental results for the existing task migration algorithm of Linux and the proposed task migration algorithm show that the proposed algorithm reduces the temperature imbalance amongst cores as compared to existing algorithm, without degrading the performance.
机译:多核架构最近被出现为当前和未来处理器的主要设计范例,然而,在同一芯片上添加多个核心并同时执行这些核心的过程增加了核心之间的温度不平衡,导致芯片增加温度又降低了处理器的可靠性。在本研究项目中,通过任务迁移,在操作系统级别地解决了多核处理器系统中的核心温度不平衡问题。所提出的任务迁移算法发现了个别任务的热贡献,并根据该热贡献来执行任务迁移,以便在核心之间保持温度平衡。所提出的任务迁移算法通过修改以C语言编写的源代码在Linux操作系统中实现。不同类别的进程(例如CPU绑定,IO绑定等)由Phoronix测试套件和Sysbench基准测试执行,以测试所提出的算法的性能。 Linux现有任务迁移算法的实验结果和所提出的任务迁移算法表明,与现有算法相比,该算法减少了核心之间的温度不平衡,而不会降低性能。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号