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

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

개선된 윔홀 라우터를 이용한 파이프라인 브로드캐스트

Pipelined Broadcast with Enhanced Wormhole Routers

  • 전민수 (고려대학교 전기공학과) ;
  • 김동승 (고려대학교 전기공학과)
  • 발행 : 2002.02.01
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

이논문은 nckdnjs 하리퍼큐브에서 O(m+n-1) 이내에서 크기 m인 메세지를 브로드캐스트 하는 파 이프라인 브로드캐스트(pipelined broadcast)를 제안한다. 이것은 도달가능집합(reachable set)으로 부터 유도된 복제 나무(replication tree)를 이용한다. 이 브로드캐스트방식은 O(m[n/log(n+1)])의 수행시간을 갖는 Ho-Kao 의 알고리즘에 비해 성능이 크게 개선된 결과이다. 브로드개스트 통신은 메시지 복제 기능을 갖는 all-port 휨홀 라우터를 이용한다. 이 논문은 알고리즘 기술 및 실제 구현시 이전 방식들과의 성능비 교도 제시한다.

This paper proposes the Pipelined Broadcast that broadcasts a message of size m in O(m+n-1) time in an n-dimensional hypercube. It is based on the replication tree, which is derived from the reachable sets. It greatly improves the performance compared to Ho-Kao s algorithm with the time of O(m[n/log(n+1)]). The communication in the broadcast uses all-port wormhole router with message replication capability. This paper includes the algorithm together with performance comparisons to previous schemes in practical implementation.

키워드

참고문헌

  1. K. Hwang and Z. Xu, Scalable parallel computing: technology, architecture, programming, McGraw Hill, 1998
  2. X. Lin and L.M. Ni, 'Deadlock-free multicast wormhole routing in multicomputer networks,' Proc. 18th Int'l Symp. Computer Architecture, pp. 116-125, 1991 https://doi.org/10.1145/115952.115965
  3. L. M. Ni and P. K. McKinley, 'A survey of wormhole routing techniques in direct networks,' IEEE Computer, vol. 26, no. 2, pp. 62-76, 1993 https://doi.org/10.1109/2.191995
  4. P. K. Mckinley and Y. Tsai, 'Collective communication in wormhole-routed massively parallel computers,' IEEE Computer, pp. 39-50, 1995 https://doi.org/10.1109/2.476198
  5. D Kim and S-H Kim, ' O(logn) numerical algorithms on a mesh with wormhole routing,' Information Processing Letters, pp. 129-136, 1994 https://doi.org/10.1016/0020-0190(94)00029-8
  6. H. Sullivan and T. R. BashkoW, 'A large scale, homogeneous, fully distributed parallel machine,' Proc. the 4th Anna Symp. Computer Architecture, vol. 5, pp. 105-124, 1977 https://doi.org/10.1145/800255.810659
  7. C.-T. Ho and M. Kao, 'Optimal broadcast in all-port wormhole-routed hypercubes,' IEEE Trans. Parallel and Distributed Systems, Vol. 6, No. 2, Feb. 1995, pp. 200-204 https://doi.org/10.1109/71.342134
  8. V. Halwan and F. Ozguner, 'Efficient multicast algorithms in all-port wormhole-routed hypercubes,' Proc. the 1997 Int'l Conf. on Parallel Processing, pp. 84-91, 1997
  9. Z. Xu and K. Hwang, 'Modeling communication overhead: MPI and MPL performance on the IBM SP2,' IEEE Transactions on Parallel and Distributed Technology, vol. 4, no. 1, 1996 https://doi.org/10.1109/88.481662
  10. Y. Tseng, S. Wang, and C. Ho, 'Efficient broadcasting in wormhole-routed multicomputers: a network-partitioning approach,' IEEE Trans. on Parallel and Distributed Systems, vol. 10, no. 1, pp. 44-61, Jan. 1999 https://doi.org/10.1109/71.744837
  11. Performance of the Cray T3E multiprocessor, http://www.sgi.com/t3e/performance.html, 1999