ARC: Adaptive Ray-tracing with CUDA, a New Ray Tracing Code for Parallel GPUs

TL;DR

This paper introduces ARC, a GPU-accelerated, photon-conserving, adaptive ray-tracing radiative transfer code that efficiently handles thousands of sources in cosmological simulations, maintaining accuracy and scalability.

Contribution

The paper presents a novel GPU-based ray-tracing algorithm for radiative transfer that is scalable, accurate, and efficient for large cosmological simulations with many sources.

Findings

Code scales linearly with the number of sources.

Performance benchmarks show linear scaling with grid size squared.

Simulation of reionization with 1024^3 grid points takes about 30 days on 64 GPUs.

Abstract

We present the methodology of a photon-conserving, spatially-adaptive, ray-tracing radiative transfer algorithm, designed to run on multiple parallel Graphic Processing Units (GPUs). Each GPU has thousands computing cores, making them ideally suited to the task of tracing independent rays. This ray-tracing implementation has speed competitive with approximate momentum methods, even with thousands of ionization sources, without sacrificing accuracy and resolution. Here, we validate our implementation with the selection of tests presented in the "cosmological radiative transfer codes comparison project," to demonstrate the correct behavior of the code. We also present a selection of benchmarks to demonstrate the performance and computational scaling of the code. As expected, our method scales linearly with the number of sources and with the square of the dimension of the 3D computational grid. Our current implementation is scalable to an arbitrary number of nodes possessing GPUs, but is limited to a uniform resolution 3D grid. Cosmological simulations of reionization with tens of thousands of radiation sources and intergalactic volumes sampled with 1024$^3$ grid points take about 30 days on 64 GPUs to reach complete reionization.