Abstract

Particle accelerators play an important role in a wide range of scientific discoveries and industrial applications. The self-consistent multi-particle simulation based on the particle-in-cell (PIC) method has been used to study charged particle beam dynamics inside those accelerators. However, the PIC simulation is time-consuming and needs to use modern parallel computers for high-resolution applications. In this paper, we implemented a parallel beam dynamics PIC code on multi-node hybrid architecture computers with multiple Graphics Processing Units (GPUs). We used two methods to parallelize the PIC code on multiple GPUs and observed that the replication method is a better choice for moderate problem size and current computer hardware while the domain decomposition method might be a better choice for large problem size and more advanced computer hardware that allows direct communications among multiple GPUs. Using the multi-node hybrid architectures at Oak Ridge Leadership Computing Facility (OLCF), the optimized GPU PIC code achieves a reasonable parallel performance and scales up to 64 GPUs with 16 million particles.

Highlights

  • The modern particle accelerator as one of the most important inventions in 20th century provides an important tool in scientific discovery and industrial application

  • We used two methods to parallelize the PIC code on multiple Graphics Processing Units (GPUs) and observed that the replication method is a better choice for moderate problem size and current computer hardware while the domain decomposition method might be a better choice for large problem size and more advanced computer hardware that allows direct communications among multiple GPUs

  • To the best of our knowledge, there was no report on the implementation of a parallel particle accelerator beam dynamics PIC code on multiple GPU nodes

Read more

Summary

Introduction

The modern particle accelerator as one of the most important inventions in 20th century provides an important tool in scientific discovery and industrial application. A number of parallel PIC beam dynamics codes using Message Passing Interface (MPI) were developed in the accelerator community for high intensity/high brightness beam simulations [2] [3] [4] [5] [6]. A number of PIC codes (especially in plasma physics community) were implemented on GPUs in previous studies and significant improvement of computing performance was reported in [15]-[27]. To the best of our knowledge, there was no report on the implementation of a parallel particle accelerator beam dynamics PIC code on multiple GPU nodes. The MPI based parallel beam dynamics PIC code, ImpactT [6], was implemented and optimized using the CUDA parallel computing platform on both a single GPU and multi-node GPU architectures.

Multi-Particle Beam Dynamics PIC Model
Implementation on Multiple GPUs
Reorder
Depositor
Poisson Solver
Particle Pushing
Performance Tests
Performance Test of the Poisson Solver
Performance Study on a Single GPU
Performance Study on Multi-Node GPUs
Findings
Conclusion
Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call