Global communications in multiprocessor simulations of flames

TL;DR

This paper examines how global communication patterns affect the performance of a parallel spectral code simulating flame dynamics, highlighting the importance of interprocessor connectivity in HPC systems.

Contribution

It provides an analysis of global communication impacts on a specific flame simulation code and discusses prospects for simulating more complex models efficiently.

Findings

Efficient simulation is possible on modern HPC systems despite global data dependencies.

Global all-to-all communication significantly influences performance.

Prospects for more sophisticated flame models are promising.

Abstract

In this paper we investigate performance of global communications in a particular parallel code. The code simulates dynamics of expansion of premixed spherical flames using an asymptotic model of Sivashinsky type and a spectral numerical algorithm. As a result, the code heavily relies on global all-to-all interprocessor communications implementing transposition of the distributed data array in which numerical solution to the problem is stored. This global data interdependence makes interprocessor connectivity of the HPC system as important as the floating-point power of the processors of which the system is built. Our experiments show that efficient numerical simulation of this particular model, with global data interdependence, on modern HPC systems is possible. Prospects of performance of more sophisticated models of flame dynamics are analysed as well.