Journal of KIISE:Computer Systems and Theory (한국정보과학회논문지:시스템및이론)

Korean Institute of Information Scientists and Engineers (한국정보과학회)
권호 표지

Torus Ring : Improving Performance of Interconnection Networks by Modifying Hierarchical Ring

Torus Ring : 계층 링 구조의 변형을 통한 상호 연결망의 성능 개선

  • 곽종욱 (서울대학교 전기컴퓨터공학부) ;
  • 반형진 (삼성전자) ;
  • 전주식 (서울대학교 전기컴퓨터공학부)
  • Published : 2005.06.01
Download PDF
⟨ Previous Next ⟩

Abstract

In multiprocessor systems, interconnection network design is critical for overall system performance. Popular interconnection networks, which are generally considered, are meshes, rings, and hierarchical rings. In this paper, we propose (')Torus Ring('), which is a modified version of hierarchical ring. Torus Ring has the same complexity as the hierarchical rings, but the only difference is the way it connects the local rings. It has an advantage over the hierarchical rings when the destination of a packet is the neighbor local ring in the reverse direction. Though the average number of hops in Torus Ring is equal to that of the hierarchical rings when assuming the uniform distribution of each transaction, the benefits of the number of hops are expected to be larger because of the spatial locality in the real environment of parallel programming. In the simulation results, latencies in the interconnection network are reduced by up to 19$\%$, and the execution times are reduced by up to 10$\%$.

다중 프로세서 시스템에서 노드 간의 연결을 제공하는 상호 연결망이 전체 시스템의 성능에서 차지하는 비중은 매우 크다 상호 연결망의 형태는 여러 종류가 있을 수 있으나 Mesh, 링, 계층 링 등의 형태가 많이 사용된다. 이 논문에서는 기존의 계층 링을 수정,한 Torus Ring을 제안한다. Torus Ring은 계층 링과 완전히 동일한 복잡도를 가지면서도 지역 링 간의 연결 방법만을 변경한 형태의 상호 연결망이다. 이 연결망은 역방향 인접 링에 대한 요청에서 홉 수의 이득을 봄으로써 평균 흡수를 감소시킨다. 또한 접근의 지역성을 고려하지 않은 균등분포의 가정 하에서도 평균 홉수의 기대값에서 계층링과 동일한 값을 가지며, 실제 병렬 프로그램이 수행되는 환경에서는 인접링에 대한 통신 비율이 증가할 가능성이 크기 때문에 더 큰 흡수의 이익을 기대할 수 있다. 이에 따라 상호 연결망의 요청과 웅답의 지연 시간이 최대 19$\%$까지 감소하였으며, 이러한 웅답 지연 시간의 단축이 수행 시간을 최대 10$\%$ 정도까지 감소시키는 결과를 가져왔다.

Keywords

References

  1. David E. Culler and Jaswinder Pal Singh with Anoop Gupta, 'Parallel Computer Architecture : A Hardware/Software Approach,' Morgan Kaufmann Publishers, Inc, 1998
  2. Sung Woo Chung, Seong Tae Jhang and Chu Shik Jhon, 'PANDA : Ring-Based Multiprocessor System using New Snooping Protocol,' In The Proceeding of ICPADS'98, pp 10-17, Dec. 1998 https://doi.org/10.1109/ICPADS.1998.741012
  3. Byoung Soon Jang, Sung Woo Chung, Seong Tae Jhang and Chu Shik Jhon, 'Efficient Schemes to Scale the Interconnection Network Bandwidth in a Ring-based Multiprocessor System,' SAC-2001(16th ACM Symposium on Applied Computing), Las Vegas, United States, pp.510-516, March 2001 https://doi.org/10.1145/372202.372447
  4. G. Ravindran and M. Stumm, 'A performance comparison of hierarchical ring- and meshconnected multiprocessor networks,' Third International Symposium on High-Performance Computer Architecture, pp. 58-69, Feb 1997
  5. G. Ravindran and M. Stumm, 'On topology and bisection bandwidth of hierarchical-ring networks for shared-memory multiprocessors,' 5th International Conference On High Performance Computing, pp. 262-269, Dec 1998 https://doi.org/10.1109/HIPC.1998.737997
  6. Z.G. Vranesic, M. Stumm, D.M. Lewis and R. White, 'Hector: a hierarchically structured shared-memory multiprocessor,' Computer, Vol.24, Iss.1, pp. 72-79, Jan. 1991 https://doi.org/10.1109/2.67196
  7. R. Grindley, T. Abdelrahman, S. Brown, S. Caranci, D. DeVries, B. Gamsa, A. Grbic, M. Gusat, R. Ho, O. Krieger, G. Lemieux, K. Loveless, N. Manjikian, P. McHardy, S. Srbljic, M. Stumm, Z. Vranesic and Z. Zilic, 'The NUMAchine multiprocessor,' Proceedings of International Conference on Parallel Processing, pp. 487-496, 2000
  8. Dongho Yoo, Inbum Jung, Seung Ryoul Maeng and Hyunglae Roh, 'Multistage ring network: a new multiple ring network for large scale multiprocessors,' Proceedings of International Workshops on Parallel Processing, pp. 290-294, 1999 https://doi.org/10.1109/ICPPW.1999.800075
  9. Guihai Chen and F.C.M. Lau, 'Shuffle-Ring: overcoming the increasing degree of hypercube,' Proceedings of Second International Symposium on High-Performance Computer Architecture, pp. 130-138, Feb 1996 https://doi.org/10.1109/HPCA.1996.501180
  10. 'Recursive cube of rings: a new topology for interconnection networks,' Y. Sun, P.Y.S. Cheung, X. Lin, IEEE Transactions on Parallel and Distributed Systems, Vol.11, Iss.3, pp, 275-286, Mar 2000 https://doi.org/10.1109/71.841743
  11. 성현중, 김형호, 장성태, 전주식, '스누핑 프로토콜을 사용하는 NUMA 시스템의 계층적 링 구조로의 확장', 정보과학회 논문지(A), pp. 1305-1317, Vol. 26, No. 11, Nov. 1999
  12. 경진미, 김인석, 김봉준, 장성태 '리피터 노드를 장착한 이중 링 CC-NUMA 시스템', 한국정보과학회:학술대회지, 2002, 10 v.2002, n.한국정보과학회 02 가을 학술발표논문집(1), pp.697-699
  13. W.J, Dally, 'Virtual-Channel Flow Control,' IEEE Trans. Parallel and Dist. Sys., vol. 3, no. 2, pp. 194-205, Mar. 1992 https://doi.org/10.1109/71.127260
  14. G. Ravindran, 'Performance Issues in the Design of Hierarchical-ring and Direct Networks for Sharedmemory Multiprocessors,' Ph.D. Dissertation, Department of Electrical and Computer Engineering, University of Toronto, Jan 1998
  15. Vijai S. Pai, Parthasarathy Ranganathan and Sarita V. Adve, 'RSIM Reference Manual,' Dept. of Electrical and Computer Engineering, Rice University, Technical Report 9705, 1997
  16. J. P. Singh, W. D. Weber, and A. Gupta, 'Splash: Stanford parallel applications for shared memory,' Tech. Rep. CSL-TR-91-469, Stanford University, 1991
  17. Steven Cameron Woo, Moriyoshi Ohara, Evan Torrie, Jaswinder Pal Singh, and Anoop Gupta, 'The SPLASH-2 Programs: Characterization and Methodological Considerations,' Proceedings of the 22nd International Symposium on Computer Architecture, 1995