期刊文献+

一种实时细颗粒度片上网络功率分配方法 被引量:1

Real-time fine-gained power budget method for network-on-chip
下载PDF
导出
摘要 片上网络(No C)不仅直接影响众核芯片的性能,而且消耗了芯片部分功率,在给定功率条件下使得片上网络的性能最优已经成为迫切需求。片上网络中路由器的工作负载(包的到达率)差异很大,需要对每一个路由器进行异构调整功率才能得到很好的性能。为此,针对如何在给定功率条件下优化No C性能进行了研究,提出了一个实时细颗粒度的功率分配方法,它能够根据每一个路由器的工作负载实时快速地分配功率,实现片上网络性能最优。实验结果表明相比其他功率分配方法,该方法平均减少26.1%的应用执行时间且具有较低的硬件开销。 Network-on-chip (NoC) not only directly influences the overall many-core performance, but also consumes a sig- nificant portion of the total chip power. How to obtain the most optimal performance under the conditions of limited power has become an urgent need. The routers' workload (the packet arrival rate) in NoC is vary difference, so needs to adjust the power for each router heterogeneously to get good performance. Thus, this paper proposed a method of real-time fine-gained power budget in NoC under the limits of power which can perceive the touters' workload to distribute the power online and quickly to obtain the most optimize performance. Experimental results show that compared several state-of-the-art methods, the proposed method can reduce 26.1% of the application execution time averagely.
出处 《计算机应用研究》 CSCD 北大核心 2016年第8期2358-2362,共5页 Application Research of Computers
基金 国家自然科学基金资助项目(61100134) 中央高校基本科研业务费专项资金资助项目(21615438)
关键词 众核芯片 片上网络 功率受限 性能最优 功率分配 many-core chip NoC power-limited optimal performance power budget
  • 相关文献

参考文献3

二级参考文献19

  • 1郭晶旭.基于快速排序的改进算法.计算机科学,2009,36(4):343-344.
  • 2Michael J.Quinn.MPI与OpenMP并行程序设计[M].陈文光,武永卫等译.北京:清华大学出版社,2004.
  • 3OpenMP标准[EB/OL].http://www.openmp.org.2010.
  • 4Driscoll J R, Gabow H N, Shrairman R, et al. Relaxed Heaps: An Alternative to Fibonacci Heaps with Applications to Parallel Computation[J]. Communications of the ACM, 1988, 31(11): 1343-1354.
  • 5Brodal G S, Tr?ff J L, Zaroliagis C D. A Parallel Priority Queue with Constant Time Operations[J]. Journal of Parallel Distributed Computing, 1998, 49(1): 4-21.
  • 6Meyer U, Sanders P. Δ-Stepping: A Parallelizable Shortest Path Algorithm[J]. Journal of Algorithms, 2003, 49(1): 114-152.
  • 7Madduri K, Bader D A, Berry J W, et al. Parallel Shortest Path Algorithms for Solving Large-scale Instances[EB/OL]. (2010-11- 21). http://www.dis.unironma1.it/~challenge9/papers.shtml.
  • 8Harish P, Narayanan P J. Accelerating Large Graph Algorithms on the GPU Using CUDA[C] //Proc. of the 14th International Conference on High Performance Computing. Berlin, Germany: Springer-Verlag, 2007.
  • 9A. Kohler and M. Radetzki, "Power management for high-performance applications on network-on-chip-based multiprocessors," in Proc. of iEEE into Canl on Cyber; Physical, and Social Computing, Beijing, 2013, pp. 77-85.
  • 10S. R. Sridhara and N. R. Shanbhag, "Coding for system-on-chip networks: a unified framework," iEEE Trans. on Very Large Scale integration Systems, vol. 13, pp. 655-667, Jun. 2005.

共引文献7

同被引文献1

引证文献1

二级引证文献2

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部