Modelling the earth's geomagnetic environment on Cray machines using PETSc and SLEPc

TL;DR

This paper enhances the British Geological Survey's geomagnetic model by replacing its eigen-solver with SLEPc on Cray machines, enabling larger system modeling and significantly improved performance through optimized data distribution and parallel computing techniques.

Contribution

It introduces a novel approach for building symmetric matrices for distributed systems and demonstrates substantial performance gains and scalability improvements using SLEPc on Cray hardware.

Findings

Achieved a 294x speed-up in the solver performance.

Supported modeling of systems with up to 100,000 coefficients.

Demonstrated effective load balancing and reduced communication in distributed matrix construction.

Abstract

The British Geological Survey's global geomagnetic model, Model of the Earth's Magnetic Environment (MEME), is an important tool for calculating the earth's magnetic field, which is continually in flux. Whilst the ability to collect data from ground based observation sites and satellites has grown, the memory bound nature of the code has proved a limitation in modelling problem sizes required by modern science. In this paper we describe work replacing the bespoke, sequential, eigen-solver with that of the SLEPc package for solving the system of normal equations. This work had a dual purpose, to break through the memory limit of the code, and thus support the modelling of much larger systems, by supporting execution on distributed machines, and to improve performance. But when adopting SLEPc it was not just the solving of the normal equations, but also fundamentally how we build and distribute the data structures. We describe an approach for building symmetric matrices in a way that provides good load balance and avoids the need for close co-ordination between processes or replication of work. We also study the memory bound nature of the code and combine detailed profiling with software cache prefetching to significantly optimise. Performance and scaling are explored on ARCHER, a Cray XC30, where we achieve a speed up for the solver of 294 times by replacing the model's bespoke approach with SLEPc. This work also provided the ability to model much larger system sizes, up to 100,000 model coefficients. Some of the challenges of modelling systems of this large scale are explored, and mitigations including hybrid MPI+OpenMP along with the use of iterative solvers are also considered. The result of this work is a modern MEME model that is not only capable of simulating problem sizes demanded by state of the art geomagnetism but also acts as further evidence to the utility of the SLEPc libary.