A High-Radix Circulant Network Topology for Efficient Collective Communication
Pages 401 - 412
Abstract
Collective communication is widely used in parallel applications. Collective-communication operations, such as Broadcast, Allreduce, and Alltoall, are frequently formed by a large number of peer-to-peer (P2P) communications. The latency of P2P communication affects the overall performance of collective communication. This paper proposes using circulant network topologies for a high-radix interconnection network to improve the performance of collective communications. The circulant network topology takes advantage of an algorithmic feature that reduces the total hop counts of collective communications. The SimGrid discrete-event simulation results showed that the execution time of the collective communication on a circulant network topology improved by 25.7% and 43.1% compared with random and dragonfly network topologies with the same degree, respectively. It also enhances 40.6% and 19.5% on average compared with 3-D torus and hypercube topologies, respectively.
References
[1]
Stunkel, C.B., et al.: The high-speed networks of the summit and sierra supercomputers. IBM J. Res. Dev. 64(3/4), 3–1 (2020)
[2]
Kim, J., Dally, W. J., Scott, S., Abts, D.: Technology-driven, highly-scalable dragonfly topology. In: ISCA, 2008, pp. 77–88 (2008)
[3]
Besta, M., Hoefler, T.: Slim fly: a cost effective low-diameter network topology. In: SC: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, pp. 348–359. IEEE (2014)
[4]
Koibuchi M, Matsutani H, Amano H, Hsu DF, and Casanova H A case for random shortcut topologies for HPC interconnects ISCA 2012 40 3 177-188
[5]
Cui K and Koibuchi M Efficient two-opt collective-communication operations on low-latency random network topologies IEICE Trans. Inf. Syst. 2020 103 12 2435-2443
[6]
Mizutani K, Yamaguchi H, Urino Y, and Koibuchi M OPTWEB: a lightweight fully connected inter-FPGA network for efficient collectives IEEE Trans. Comput. 2021 70 6 849-862
[7]
Bruck J, Ho C-T, Kipnis S, Upfal E, and Weathersby D Efficient algorithms for all-to-all communications in multiport message-passing systems IEEE Trans. Parallel Distrib. Syst. 1997 8 11 1143-1156
[8]
Liu, M.T.: Distributed loop computer networks. Adv. Comput. 17, 163–221. Elsevier (1978)
[9]
Bermond J-C, Comellas F, and Hsu DF Distributed loop computer-networks: a survey J. Parallel Distrib. Comput. 1995 24 1 2-10
[10]
Junginger, M., Lee, Y.: The multi-ring topology-high-performance group communication in peer-to-peer networks. In: Second International Conference on Peer-to-Peer Computing, 2002, pp. 49–56 (2002)
[11]
Park, J.-H., Chwa, K.-Y.: Recursive circulant: a new topology for multicomputer networks. In: International Symposium on Parallel Architectures, Algorithms and Networks (ISPAN), 1994, pp. 73–80 (1994)
[12]
Tang S-M, Wang Y-L, and Li C-Y Generalized recursive circulant graphs IEEE Trans. Parallel Distrib. Syst. 2011 23 1 87-93
[13]
Huang, X., Ramos, A.F., Deng, Y.: Optimal circulant graphs as low-latency network topologies, arXiv preprint arXiv:2201.01342 (2022)
[14]
Chunduri, S., Parker, S., Balaji, P., Harms, K., Kumaran, K.: Characterization of MPI usage on a production supercomputer. In: SC: International Conference for High Performance Computing, pp. 386–400. Storage and Analysis, Networking (2018)
[15]
Open MPI: Open Source High Performance Computing. http://www.open-mpi.org/
[16]
MPICH | High-Performance Portable MPI. http://www.mpich.org/
[17]
[18]
Thakur R, Rabenseifner R, and Gropp W Optimization of collective communication operations in MPICH Int. J. High Perform. Comput. Appl. 2005 19 1 49-66
[19]
Boesch F and Tindell R Circulants and their connectivities J. Gr. Theory 1984 8 4 487-499
[20]
Casanova H, Giersch A, Legrand A, Quinson M, and Suter F Versatile, scalable, and accurate simulation of distributed applications and platforms J. Parallel Distrib. Comput. 2014 74 10 2899-2917
[21]
Bertsekas DP, Özveren C, Stamoulis GD, Tseng P, and Tsitsiklis JN Optimal communication algorithms for hypercubes J. Parallel Distrib. Comput. 1991 11 4 263-275
[22]
Ho C-T and Kao M-Y Optimal broadcast in all-port wormhole-routed hypercubes IEEE Trans. Parallel Distrib. Syst. 1995 6 2 200-204
Recommendations
Collective Communication in Recursive Dual-Net: An Interconnection Network for High-Performance Parallel Computer Systems of the Next Generation
CIT '10: Proceedings of the 2010 10th IEEE International Conference on Computer and Information TechnologyIn this paper, we propose efficient routing algorithms for collective communication in a newly proposed, versatile network, called a recursive dual-net (RDN). The RDN can be used as a candidate for the interconnection of supercomputers of the next ...
Multidestination Message Passing in Wormhole k-ary n-cube Networks with Base Routing Conformed Paths
This paper proposes multidestination message passing on wormhole k-ary n-cube networks using a new base-routing-conformed-path (BRCP) model. This model allows both unicast (single-destination) and multidestination messages to co-exist in a given network ...
Broadcast communication operations for hyper hexa-cell interconnection network
The hyper hexa-cell (HHC) interconnection network is relatively a new interconnection network, which is constructed from hexa-cell and hypercube topologies. HHC is developed to support parallel algorithms for solving computation and communication ...
Comments
Information & Contributors
Information
Published In
Dec 2022
525 pages
ISBN:978-3-031-29926-1
DOI:10.1007/978-3-031-29927-8
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.
Publisher
Springer-Verlag
Berlin, Heidelberg
Publication History
Published: 08 April 2023
Author Tags
Qualifiers
- Article
Contributors
Other Metrics
Bibliometrics & Citations
Bibliometrics
Article Metrics
- 0Total Citations
- 0Total Downloads
- Downloads (Last 12 months)0
- Downloads (Last 6 weeks)0
Reflects downloads up to 23 Dec 2024