An Efficient Method for Determining Full Point-to-Point Latency of Arbitrary Indirect HPC Networks
Point-to-point latency is one of the most important metrics for high performance computer networks and is used widely in communication performance modeling, link-failure detection, and application optimization. However, it is often hard to determine the full-scale point-to-point latency of large scale HPC networks since it often requires measurements to the square of the number of terminal nodes. In this paper, we propose an efficient method to generate measurement plans for arbitrary indirect HPC networks and reduces the measurement requirements from \(O(n^2)\) to m, which is often O(n) in modern indirect networks containing n nodes and m links, thus significantly reduces the latency measure overhead. Both analysis and experiments show that the proposed method can reduce the overhead of large-scale fat-tree networks by orders of magnitudes.
This work in this paper is supported by the National Key R&D Program of China under Grant NO. 2018YFB0203901, Science Challenge Project, NO. TZ2016002, and the National Natural Science Foundation of China under Grant No. 61772053. The authors would like to thank the reviewers for their valuable comments.
- 5.Sen, S., Wang, J.: Analyzing peer-to-peer traffic across large networks. In: Proceedings of the 2nd ACM SIGCOMM Workshop on Internet measurement, no. 2, pp. 137–150 (2002)Google Scholar
- 9.Lin, X,Y., Chung, Y,C., Huang, T,Y.: A multiple LID routing scheme for fat-tree-based InfiniBand networks. In: Parallel and Distributed Processing Symposium, 18, p. 11 (2004)Google Scholar
- 10.Hoefler, T., Schneider, T., Lumsdaine, A.: LogGOPSim: simulating large-scale applications in the LogGOPS model. In: Proceedings of ACM International Symposium on High Performance Distributed Computing, 19, pp. 597–604 (2010)Google Scholar