Enabling GPU Accelerated Computing in the SUNDIALS Time Integration Library

TL;DR

This paper discusses the development of GPU-accelerated features in the SUNDIALS library, enabling efficient time integration for large-scale scientific applications on modern supercomputers with heterogeneous architectures.

Contribution

It introduces new GPU-enabled data structures, programming support, and utilities in SUNDIALS, along with performance evaluations on Summit and Frontier supercomputers.

Findings

Negligible performance overhead from new infrastructure

Significant speedups on NVIDIA and AMD GPUs

Successful deployment on Summit and Frontier supercomputers

Abstract

As part of the Exascale Computing Project (ECP), a recent focus of development efforts for the SUite of Nonlinear and DIfferential/ALgebraic equation Solvers (SUNDIALS) has been to enable GPU-accelerated time integration in scientific applications at extreme scales. This effort has resulted in several new GPU-enabled implementations of core SUNDIALS data structures, support for programming paradigms which are aware of the heterogeneous architectures, and the introduction of utilities to provide new points of flexibility. In this paper, we discuss our considerations, both internal and external, when designing these new features and present the features themselves. We also present performance results for several of the features on the Summit supercomputer and early access hardware for the Frontier supercomputer, which demonstrate negligible performance overhead resulting from the additional infrastructure and significant speedups when using both NVIDIA and AMD GPUs.