Complex network metrics as indicators of scalability of parallel programs
Abstract
This work presents a study on potential scalability indicators of parallel programs to be defined by methodological analysis, based on metrics of complex dynamic networks and time series. Communication events are extracted from an execution trace and a communication graph is generated in order to extract metrics from complex networks. The metrics are then presented on a graph over a period of time, resulting in a time series, where statistical measures can be applied. The objective is to characterize the parallel programs from these time series so that scalability indicators related to the communication pattern can be obtained.
Keywords:
complex networks, scalability, prediction, graphs
References
Bailey, D., Harris, T., Saphir, W., Van Der Wijngaart, R., Woo, A., and Yarrow, M. (1995). The nas parallel benchmarks 2.0. Technical report, Technical Report NAS-95-020, NASA Ames Research Center.
Cressie, N. and Wikle, C. K. (1998). The variance-based cross-variogram: You can add apples and oranges. Mathematical Geology, 30(7):789–799.
Hayashi, T. and Yoshida, N. (2005). On covariance estimation of non-synchronously observed diffusion processes. Bernoulli, 11(2):359–379.
Schult, D. A. (2008). Exploring network structure, dynamics, and function using networkx. In In Proceedings of the 7th Python in Science Conference (SciPy), pages 11–15.
Score-P Developer Community (2019). Scalable performance measurement infrastructure for parallel codes (Score-P).
Snir, M., Otto, S., Huss-Lederman, S., Walker, D., and Dongarra, J. (1998). MPI-The Complete Reference, Volume 1: The MPI Core. MIT Press, Cambridge, MA, USA, 2nd. (revised) edition.
Stringhini, D. and Fazenda, A. (2015). Characterizing communication patterns of parallel programs through graph visualization and analysis. pages 565–576.
Cressie, N. and Wikle, C. K. (1998). The variance-based cross-variogram: You can add apples and oranges. Mathematical Geology, 30(7):789–799.
Hayashi, T. and Yoshida, N. (2005). On covariance estimation of non-synchronously observed diffusion processes. Bernoulli, 11(2):359–379.
Schult, D. A. (2008). Exploring network structure, dynamics, and function using networkx. In In Proceedings of the 7th Python in Science Conference (SciPy), pages 11–15.
Score-P Developer Community (2019). Scalable performance measurement infrastructure for parallel codes (Score-P).
Snir, M., Otto, S., Huss-Lederman, S., Walker, D., and Dongarra, J. (1998). MPI-The Complete Reference, Volume 1: The MPI Core. MIT Press, Cambridge, MA, USA, 2nd. (revised) edition.
Stringhini, D. and Fazenda, A. (2015). Characterizing communication patterns of parallel programs through graph visualization and analysis. pages 565–576.
Published
2020-08-19
How to Cite
MAGUETA, Marcos Paulo Bispo; STRINGHINI, Denise.
Complex network metrics as indicators of scalability of parallel programs. In: REGIONAL SCHOOL OF HIGH PERFORMANCE COMPUTING FROM SÃO PAULO (ERAD-SP), 11. , 2020, Evento Online.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2020
.
p. 42-45.
DOI: https://doi.org/10.5753/eradsp.2020.16882.
