《计算机应用研究》|Application Research of Computers

基于温度感知任务调度的3D NoC混合拓扑结构

3D NoC hybrid topology based on thermal-aware task migration

免费全文下载 (已被下载 次)  
获取PDF全文
作者 冯申杰,李冰,程良伦
机构 1.广东工业大学 计算机学院,广州 510006;2.华南理工大学 软件学院,广州 510006
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2017)08-2395-04
DOI 10.3969/j.issn.1001-3695.2017.08.036
摘要 3D NoC较高的功率密度容易造成温度过高,对系统性能和芯片可靠性造成负面影响。利用温度感知任务调度来控制节点温度的思路是在运行时把“热”节点上的任务迁移到“冷”节点上,这不可避免会出现迁移之后任务间通信距离变大进而影响整体性能。因此,在任务调度的过程中保持通信开销已经成为迫切需求。提出了分层次的ring/mesh 混合拓扑结构RMH,可以在任务迁移的同时保持原来较小的通信延迟。仿真结果表明,相比于3D NoC拓扑结构,RMH拓扑可以有效缓解散热问题,并且平均减少31.1%的网络延迟。
关键词 众核芯片;片上网络;任务调度;三维混合拓扑
基金项目 广东省战略性新兴产业发展专项资助项目(2012A09100013)
广东省联合基金重点资助项目(U2012A002D01)
本文URL http://www.arocmag.com/article/01-2017-08-036.html
英文标题 3D NoC hybrid topology based on thermal-aware task migration
作者英文名 Feng Shenjie, Li Bing, Cheng Lianglun
机构英文名 1.SchoolofComputer,GuangdongUniversityofTechnology,Guangzhou510006,China;2.SchoolofSoftware,SouthChinaUniversityofTechnology,Guangzhou510006,China
英文摘要 High power density of 3D NoC system is easy to cause on chip high temperature problem, which has a destructive effect on the system performance and chip reliability. Thermal aware task scheduling method mean to migrate hot task to relative cool task, but this way will increase the communication distance between the two node, and then has a destructive affection of overall performance. Therefor it needs to keep the original low communication overhead when using task migration met-hod. This paper proposed a hierarchical ring/mesh hybrid 3D NoC topology: RMH. This topology could balance the communication need and thermal need of 3D NoC system. RMH could keep original communication overhead when tasks migrated. The simulation results show that, compared to the traditional topology, the RMH topology can effectively alleviate the thermal problem, and reduces 31.1% of network latency on average.
英文关键词 many-core chip; network on chip; task migration; 3D hybrid topology
参考文献 查看稿件参考文献
  [1] 张大坤, 黄翠, 宋国治. 三维片上网络研究综述[J] . 软件学报, 2016, 27(1):155-187.
[2] Hassanpour N, Hessabi S, Hamedani P K. Temperature control in three-network on chips using task migration[J] . LET Computers & Digital Techniques, 2013, 7(6):274-281.
[3] Fourmigue A, Beltrame G, Nicolescu G, et al. Multi-granularity thermal evaluation of 3D MPSoC architectures[C] //Proc of Conference on Design, Automation & Test in Europe. 2011:1-4.
[4] Ausavarungnirun R, Fallin C, Yu Xiangyao, et al. Achieving both high energy efficiency and high performance in on-chip communication using hierarchical rings with deflection routing[J/OL] . (2016-02-18). http://arxiv. crg/abs/1602. 02005.
[5] Hamedani P K, Hessabi S, Sarbazi-Azad H, et al. Exploration of temperature constraints for thermal aware mapping of 3D networks on chip[J] . International Journal of Adaptive Resilient & Autonomic Systems, 2012, 4(3):499-506.
[6] Chou Chenling, Ogras U Y, Marculescu R. Energy-and performance-aware incremental mapping for networks on chip with multiple voltage levels[J] . IEEE Trans on Computer-Aided Design of Integrated Circuits and Systems, 2008, 27(10):1866-1879.
[7] Al Faruque M A, Krist R, Henkel J. ADAM:run-time agent-based distributed application mapping for on-chip communication[C] //Proc of Design Automation Conference. 2008:760-765.
[8] Feero B S, Pande P P. Networks-on-chip in a three-dimensional environment:a performance evaluation[J] . IEEE Trans on Computers, 2008, 58(1):32-45.
[9] Chen Sun, Chen C H O, Kurian G, et al. DSENT:a tool connecting emerging photonics with electronics for opto-electronic networks-on-chip modeling[C] //Proc of the 6th IEEE/ACM International Symposium on Networks on Chip. 2012:201 - 210.
[10] 饶东宁, 李建华. 一种实时细颗粒度片上网络功率分配方法[J] . 计算机应用研究, 2016, 33(8):2358-2362.
[11] Zhang Wang, Hou Ligang, Wang Jinhui, et al. Comparison research between XY and odd-even routing algorithm of a 2-dimension 3×3 mesh topology network-on-chip[C] //Proc of WRI Global Congress on Intelligent Systems. 2009:329-333.
收稿日期 2016/6/5
修回日期 2016/7/26
页码 2395-2398
中图分类号 TP303
文献标志码 A