BDDC Preconditioning on GPUs for Cardiac Simulations

TL;DR

This paper presents a GPU-accelerated BDDC preconditioner and solver for cardiac electrophysiology simulations, achieving high scalability and efficiency on large GPU clusters, advancing computational modeling of diseased hearts.

Contribution

It introduces a portable, efficient BDDC preconditioner implementation optimized for GPUs, enabling scalable cardiac simulations with high performance.

Findings

Achieved over 90% scalability on 100 GPUs

Developed a portable GPU-based BDDC solver

Enhanced modeling of cardiac microstructure

Abstract

In order to understand cardiac arrhythmia, computer models for electrophysiology are essential. In the EuroHPC MicroCARD project, we adapt the current models and leverage modern computing resources to model diseased hearts and their microstructure accurately. Towards this objective, we develop a portable, highly efficient, and performing BDDC preconditioner and solver implementation, demonstrating scalability with over 90% efficiency on up to 100 GPUs.