Performance-portable GPU acceleration of the hybrid particle-in-cell code dHybridR
Abstract
Hybrid particle-in-cell simulations are widely used to study kinetic processes in collisionless astrophysical and space plasmas, yet the high computational cost of large-scale three-dimensional runs has largely confined production studies to two dimensions or restricted domains. To address this challenge, we present a performance-portable GPU implementation of the hybrid particle-in-cell code dHybridR. The implementation combines OpenMP target offloading with specialized SYCL kernels for the most computationally expensive operations, while preserving a unified CPU-GPU codebase that supports Intel, AMD, and NVIDIA GPUs. On the exascale supercomputers Aurora and Frontier, dHybridR achieves weak scaling efficiencies of 86% to 97% across 49,152 accelerators, and at 256 particles per cell, its full-node GPU throughput is up to 32× that of the vectorized CPU implementation at approximately 90% less energy per particle-update. To our knowledge, no other hybrid particle-in-cell code has reported GPU performance at this scale, leaving dHybridR uniquely positioned to exploit exascale systems. These advances substantially lower the computational barrier to large-scale three-dimensional hybrid-kinetic simulations of collisionless plasmas.
Related
- Wave-particle acceleration during supernova explosions
- Nonthermal Particle Acceleration by Magnetic Pumping in Pulsating Plasmas
- Particle Acceleration in Alfvénic Turbulence with a Strong Guide Field
- Kink instability as a particle acceleration mechanism in X-ray binaries and the connection to the outburst cycle
- Guide field effects on particle acceleration in three-dimensional relativistic magnetic reconnection
BibTeX
@misc{ostler2026performanceportable,
title = {{Performance-portable GPU acceleration of the hybrid particle-in-cell code dHybridR}},
author = {Bricker Ostler and Miha Cernetic and Damiano Caprioli},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.28422},
eprint = {2609.28422},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.28422},
}
RIS
TY - UNPB
TI - Performance-portable GPU acceleration of the hybrid particle-in-cell code dHybridR
AU - Bricker Ostler
AU - Miha Cernetic
AU - Damiano Caprioli
PY - 2026
DA - 2026/09/23
PB - arXiv
DO - 10.48550/arxiv.2609.28422
UR - https://arxiv.org/abs/2609.28422
AB - Hybrid particle-in-cell simulations are widely used to study kinetic processes in collisionless astrophysical and space plasmas, yet the high computational cost of large-scale three-dimensional runs has largely confined production studies to two dimensions or restricted domains. To address this challenge, we present a performance-portable GPU implementation of the hybrid particle-in-cell code dHybridR. The implementation combines OpenMP target offloading with specialized SYCL kernels for the most computationally expensive operations, while preserving a unified CPU-GPU codebase that supports Intel, AMD, and NVIDIA GPUs. On the exascale supercomputers Aurora and Frontier, dHybridR achieves weak scaling efficiencies of 86% to 97% across 49,152 accelerators, and at 256 particles per cell, its full-node GPU throughput is up to 32× that of the vectorized CPU implementation at approximately 90% less energy per particle-update. To our knowledge, no other hybrid particle-in-cell code has reported GPU performance at this scale, leaving dHybridR uniquely positioned to exploit exascale systems. These advances substantially lower the computational barrier to large-scale three-dimensional hybrid-kinetic simulations of collisionless plasmas.
ER -