GPU solver speeds up full 3D tokamak electromagnetic simulations
TokaGLINT: A Scalable GPU-Tailored Implicit Solver for Full 3D Tokamak Electromagnetic Simulations
Distributed, Parallel, and Cluster Computing
Summary
Simulating the electromagnetic fields inside a tokamak – a device to contain hot plasma for fusion energy – is very computationally intensive. The authors developed TokaGLINT, a software tool that runs on graphics processing units (GPUs) to perform these simulations faster and more efficiently. It uses clever ways to split the problem into smaller pieces and speed up calculations while keeping the results stable over time. This lets scientists simulate long-duration plasma behaviors in 3D with high accuracy.
What this means in practice
- •For plasma physics simulation teams: Run long-duration, high-fidelity 3D simulations of tokamak electromagnetic fields faster using GPUs for fusion research.
- •For high performance computing engineers: Design scalable GPU-accelerated solvers for large 3D electromagnetic problems that maintain numerical stability over long times.
Authors
Zifan Yang, Haoyuan Zhang, Jialin Li, Wu Yuan, Xiazhen Liu, Jian Zhang, Jianyuan Xiao, Shan Liang
Abstract
We introduce TokaGLINT, a GPU-accelerated implicit solver for electromagnetic field computations in full 3D tokamak simulations, aimed at efficient large-scale parallel GPU computing. Its central innovation lies in the co-design of hierarchical domain decomposition and a fast exact local solver, where hierarchical partitioning is tailored to match fine-grained intra-card subdomains and exploit the tensor-based solver dedicated to curvilinear-coordinate symplectic CN-FDTD-discretized 3D Maxwell equations. Backed by automated operator fusion and batching customized for the intra-card multi-subdomain structure, the solver decouples unknowns through discrete transformations and leverages tensor-structured computations to achieve high hardware utilization, while preserving the long-time stability characteristic of symplectic discretizations. TokaGLINT scales the electromagnetic field solve beyond 10,000 GPUs, achieving 90.1% weak and 53.9% strong scaling efficiency, while delivering a 2.67X single-node speedup over an unpreconditioned BiCGStab baseline (HIP-enabled HYPRE). It is validated in EAST tokamak simulations within the SymPIC plasma simulation code, enabling high-fidelity long-duration modeling.