A fully parallel, high precision, N-body code running on hybrid computing platforms

TL;DR

This paper introduces HiGPUs, a fully parallel high-precision N-body simulation code that leverages hybrid CPU and GPU computing to efficiently simulate systems with up to 8 million bodies, enabling new large-scale astrophysical studies.

Contribution

The paper presents a novel, fully parallel implementation of a high-order Hermite N-body integrator utilizing both CPU and GPU resources, achieving unprecedented scalability and accuracy.

Findings

Successfully simulated 8 million bodies on a supercomputer cluster

Achieved high accuracy with Hermite's integration scheme

Demonstrated efficient parallelization on hybrid CPU-GPU platforms

Abstract

We present a new implementation of the numerical integration of the classical, gravitational, N-body problem based on a high order Hermite's integration scheme with block time steps, with a direct evaluation of the particle-particle forces. The main innovation of this code (called HiGPUs) is its full parallelization, exploiting both OpenMP and MPI in the use of the multicore Central Processing Units as well as either Compute Unified Device Architecture (CUDA) or OpenCL for the hosted Graphic Processing Units. We tested both performance and accuracy of the code using up to 256 GPUs in the supercomputer IBM iDataPlex DX360M3 Linux Infiniband Cluster provided by the italian supercomputing consortium CINECA, for values of N up to 8 millions. We were able to follow the evolution of a system of 8 million bodies for few crossing times, task previously unreached by direct summation codes. The code is freely available to the scientific community.