Tuning Spectral Element Preconditioners for Parallel Scalability on GPUs

TL;DR

This paper evaluates the parallel scalability of spectral element preconditioners on GPUs, focusing on Chebyshev-accelerated Schwarz methods and their effectiveness in large-scale Navier-Stokes pressure solves.

Contribution

It introduces a detailed parallel scaling analysis of Chebyshev-accelerated Schwarz preconditioners on GPUs, demonstrating their robustness and efficiency for spectral element discretizations.

Findings

Chebyshev-accelerated Schwarz methods are effective preconditioners on GPUs.

Performance improves with automated run-time preconditioner selection.

Scalability is demonstrated across a wide range of processor counts.

Abstract

The Poisson pressure solve resulting from the spectral element discretization of the incompressible Navier-Stokes equation requires fast, robust, and scalable preconditioning. In the current work, a parallel scaling study of Chebyshev-accelerated Schwarz and Jacobi preconditioning schemes is presented, with special focus on GPU architectures, such as OLCF's Summit. Convergence properties of the Chebyshev-accelerated schemes are compared with alternative methods, such as low-order preconditioners combined with algebraic multigrid. Performance and scalability results are presented for a variety of preconditioner and solver settings. The authors demonstrate that Chebyshev-accelerated-Schwarz methods provide a robust and effective smoothing strategy when using $p$-multigrid as a preconditioner in a Krylov-subspace projector. The variety of cases to be addressed, on a wide range of processor counts, suggests that performance can be enhanced by automated run-time selection of the preconditioner and associated parameters.