Simulating MPI applications: the SMPI approach

Augustin Degomme (1) , Arnaud Legrand (2) , Georges Markomanolis (3) , Martin Quinson (4) , Mark Lee Stillwell (5) , Frédéric Suter (6, 7)

1 Unibas - Université de Bâle = University of Basel = Basel Universität
2 POLARIS - Performance analysis and optimization of LARge Infrastructures and Systems
3 KAUST - King Abdullah University of Science and Technology [Saudi Arabia]
4 MYRIADS - Design and Implementation of Autonomous Distributed Systems
5 Imperial College London
6 AVALON - Algorithms and Software Architectures for Distributed and HPC Platforms
7 CC-IN2P3 - Centre de Calcul de l'IN2P3

This article summarizes our recent work and developments on SMPI, a flexible simulator of MPI applications. In this tool,we took a particular care to ensure our simulator could be used to produce fast and accurate predictions in a wide variety of situations.Although we did build SMPI on SimGrid whose speed and accuracy had already been assessed in other contexts, moving suchtechniques to a HPC workload required significant additional effort. Obviously, an accurate modeling of communications and networktopology was one of the key to such achievements. Another less obvious key was the choice to combine in a single tool the possibilityto do both offline and online simulation.

Mots clés: MPI runtime and applications; Performance prediction and extrapolation; Simulation

Domaines: Calcul parallèle, distribué et partagé [cs.DC]; Modélisation et simulation

Augustin Degomme, Arnaud Legrand, Georges Markomanolis, Martin Quinson, Mark Lee Stillwell, et al.. Simulating MPI applications: the SMPI approach. IEEE Transactions on Parallel and Distributed Systems, 2017, 28 (8), pp.14. ⟨10.1109/TPDS.2017.2669305⟩. ⟨hal-01415484v2⟩