From MPI to MPI+OpenACC: Conversion of a legacy FORTRAN PCG solver for the spherical Laplace equation

TL;DR

This paper discusses converting a legacy MPI FORTRAN solver for the spherical Laplace equation to include OpenACC, enabling multi-GPU acceleration for space weather modeling, with performance analysis on multi-node systems.

Contribution

It presents a practical approach and insights for integrating OpenACC into legacy MPI FORTRAN codes for GPU acceleration.

Findings

MPI+OpenACC achieves significant speedups on multi-GPU systems.

Key strategies and challenges in code conversion are highlighted.

Performance results demonstrate improved scalability and efficiency.

Abstract

A real-world example of adding OpenACC to a legacy MPI FORTRAN Preconditioned Conjugate Gradient code is described, and timing results for multi-node multi-GPU runs are shown. The code is used to obtain three-dimensional spherical solutions to the Laplace equation. Its application is finding potential field solutions of the solar corona, a useful tool in space weather modeling. We highlight key tips, strategies, and challenges faced when adding OpenACC. Performance results are shown for running the code with MPI-only on multiple CPUs, and with MPI+OpenACC on multiple GPUs and CPUs.