GPU-accelerated Linear Algebra for Coupled Solvers in Industrial CFD Applications with OpenFOAM

TL;DR

This paper presents the development and application of GPU-accelerated coupled linear algebra solvers for industrial CFD problems, demonstrating significant speedups and improved stability over CPU-based methods in real-world test cases.

Contribution

It introduces heterogeneous GPGPU implicit CFD coupled solvers using external libraries, achieving high performance and stability improvements in industrial applications.

Findings

NASA CRM case speedup >4x on GPU

DriveAER case shows better stability and reduced time

Successful deployment on Leonardo Labs supercomputer

Abstract

The present work describes the development of heterogeneous GPGPU implicit CFD coupled solvers, encompassing both density- and pressure- based approaches. In this setup, the assembled linear matrix is offloaded onto multiple GPUs using specialized external libraries to solve the linear problem efficiently. These coupled solvers are applied to two industrial test cases representing common scenarios: the NASA CRM in a transonic regime and the external aerodynamics study of the DriveAER car. Significant performance enhancements are evident when compared to their CPU counterparts. Specifically, the NASA CRM case achieves an overall speedup of more than 4x, while the DriveAER test case demonstrates improved stability and reduced computational time compared to segregated solvers. All calculations were carried out utilizing the GPU-based partition of the davinci-1 supercomputer at the Leonardo Labs, featuring 82 GPU-accelerated nodes.