期刊文献+

基于FPGA的高性能计算中全局流水的研究 被引量:2

Research of pipeline with FPGA in HPC
下载PDF
导出
摘要 针对FPGA的全局流水进行了研究,采用CPU+FPGA的混合架构,论证了FPGA实现全局流水的优越性:使用FPGA进行全局流水可以在CPU处理过程中减少FPGA等待时间,提高FPGA的利用率;可以减少FPGA与CPU之间的通信量以及程序在CPU端的存储开销;可以均衡CPU负载,使得CPU有空闲时间处理其它任务。用N-Body的FMM算法作为例子,对优越性分别作了分析,并设计了实验方案,实验结果表明了FPGA实现全局流水的优越性。 The pipeline with FPGA is studied, the advantage of the pipeline with FPGA: reduction on the spare time of FPGA whenCPU processing, decreasing amounts of communication between FPGA and CPU, balancing the load of CPU in order to save the power ofCPU. An experiment scheme with FMM algorithm is designed and run on CPU+FPGA heterogeneous architecture. Finally, the result of experiments prove the advantage of pipeline with FPGA.
出处 《计算机工程与设计》 CSCD 北大核心 2011年第10期3382-3385,3390,共5页 Computer Engineering and Design
基金 上海市重点学科建设基金项目(J50103)
关键词 全局流水 异构体系结构 现场可编程门阵列 多体问题 快速多极算法 pipeline heterogeneous architecture FPGA N-body FMM
  • 相关文献

参考文献15

  • 1Wisniewski, Remigiusz. Synthesis of compositional micropro- gram control units for programmable devices[M].Zielona Gora: University of Zielona Gora,ISBN 978-83-7481-293-1,2009.
  • 2Buyukkurt B,Najjar W. Compiler generated systolic arrays for wavefront algorithm acceleration on FPGAs [C]. International Conference on Field Programmable Logic and Applications, 2008:655-658.
  • 3Kindratenko Volodymyr, Wilhelmson Robert.High performance computing with accelerators[J].IEEE Computing in Science & Engineering,2010,12(4): 12-16.
  • 4Nawaz,Marconi,Bertels,et al.Flexible pipelining design for re- cursive variable expansion [C]. IEEE International Symposium on Parallel & Distributed Processing,2009:1-8.
  • 5Storaasli O, Yu W, Strenski D, et al. Performance evaluation of FPGA-based biological applications [C]. Corvallis, OR, USA: Cray User Group Inc,2007.
  • 6Bondhugula U,Devulapalli A,Dinan J,et al.Hardware/software integration for all-pairs shortest-paths on a reconfigurable super- computer[C].Proc 14th IEEE Syrup Field-Programmable Cus- tom Computing Machines,2006.
  • 7Hagihara Y, Celestial mechanics [M]. Cambridge: MIT Press, 1970-1976.http://baike.qiji.cn/Detailedd7057.html.
  • 8Greengard L,Rokhlin V.A fast algorithm for particle simula- tions [J]. Journal of Computational Physics, 1987,73 (2): 325- 348.
  • 9Felipe A Cruz,Matthew G Knepley, Barba L A.Petfrnm-a dyna- mically load balancing parallel fast multipole library[DB/OL]. http://arxiv.org/abs/0905.2637,2009-05-15/2010-07-20.
  • 10Barnes J,Hut E A hierarchical O(N log N) force-calculation al- gorithm[J] .Nature, 1986,324(6096) :446-449.

同被引文献20

  • 1张树刚,张遂南,黄士坦.CRC校验码并行计算的FPGA实现[J].计算机技术与发展,2007,17(2):56-58. 被引量:43
  • 2Prasanna Sundararajan. High Performance Computing Using FPGAs XILINX White Paper[ OL]. WP375,2010.
  • 3Dimond Rob, Racanière Srbastien, Pell Oliver. Accelerating Large- Scale HPC Applications Using FPGAs[ C]//IEEE 2011. Germany : Proceedings - 2011 20th Symposium on Computer Arithmetic ,2011 : 191 - 192.
  • 4罗兴国,等.PRCA:一种高效能计算体系结构[C]//2012高效能计算机体系结构国际高端论坛,上海,2012,10.
  • 5Xilinx. Virtex-5 Family Overview. Xilinx Product Specification DS100 [OL]. http ://www. xilinx. com/2012.
  • 6Xilinx. 7 Series FPGAs Overview. Xilinx Advance Product Specification DS180[OL]. http://www. xilinx. com/2012.
  • 7John Hennessy,David Patterson. Computer Architecture: A Quantita- tive Approach[ M ]. 4th ed. Morgan Kaufmann,2006.
  • 8Zhe Zheng, Yongxin Zhu, Xu Wang, et al. Revealing Feasibility of FMM on ASIC: Efficient Implementation of N-Body Problem on FPGA [ C ]//IEEE International Conference on Computational Science and Engineering, Hang Kong. 2010 : 132 - 139.
  • 9陈金平,王生泽,吴文英.基于LabVIEW的串口通信数据校验和的实现方法[J].自动化仪表,2008,29(3):32-34. 被引量:20
  • 10余学涛,孔雪,王绪,祝永新,何卫锋,倪明,谢光伟,雷咏梅,单健晨.FMM能效分析及其ASIC可行性评估[J].计算机工程,2011,37(13):265-268. 被引量:1

引证文献2

二级引证文献5

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