Shared Memory Pipelined Parareal

TL;DR

This paper presents an OpenMP implementation of pipelined Parareal, demonstrating comparable runtime to MPI versions but with energy savings and reduced memory usage, while noting increased implementation complexity.

Contribution

It introduces a pipelined OpenMP version of Parareal and compares its performance and resource consumption to MPI-based methods.

Findings

OpenMP version has similar runtime to MPI version.

OpenMP consumes about 7% less energy.

OpenMP has a smaller memory footprint.

Abstract

For the parallel-in-time integration method Parareal, pipelining can be used to hide some of the cost of the serial correction step and improve its efficiency. The paper introduces a basic OpenMP implementation of pipelined Parareal and compares it to a standard MPI-based variant. Both versions yield almost identical runtimes, but, depending on the compiler, the OpenMP variant consumes about 7% less energy and has a significantly smaller memory footprint. However, its higher implementation complexity might make it difficult to use in legacy codes and in combination with spatial parallelisation.