GPU acceleration of optical photon propagation in low photon yield applications: Opticks for the Electron Ion Collider

Abstract

The bulk of the time required to simulate Cherenkov and scintillation detectors is consumed by optical photon transport, a task that GPUs are uniquely qualified to perform. We present EIC-Opticks, a fork of Opticks, which uses event aggregation to drastically accelerate photon transport simulation for low-to-moderate photon yield experiments. During the full Geant4 Monte Carlo simulation of a given detector, optical photon simulation is performed on GPU(s) using the NVIDIA OptiX framework. We validate this approach using the ePIC pfRICH detector. We find GPU and CPU simulations in excellent agreement, with hit totals agreeing at the

level and wavelength/radial projections giving

$$\chi ^2/\textrm{ndf}\simeq 1$$ χ 2 / ndf ≃ 1

. At the pfRICH physics program benchmark of

$$10^5$$ 10 5

muons at

$$p=5~\textrm{GeV}/c$$ p = 5 GeV / c

, EIC-Opticks shows

$$13.2\pm 0.5\times $$ 13.2 ± 0.5 ×

speedup over 20-thread Geant4 and

$$234\pm 10\times $$ 234 ± 10 ×

over a single thread. In the case of low-to-moderate applications event aggregation reduces the per-photon simulation time from about

$$40~\upmu \textrm{s}$$ 40 μ s

for single events to about

$$20~\textrm{ns}$$ 20 ns

with batching, a factor of about

$$2.3\times 10^3$$ 2.3 × 10 3

. In order to make EIC-Opticks easily installable, we authored a Spack package that makes it possible to install it with a single command. Additionally, a Docker image is provided for users with EIC-Opticks installed. EIC-Opticks provides guardrails for common pitfalls such as deeply nested boolean volumes, maximum-reflection scans, and OptiX ray-origin-offset scans.