Geant4 performance optimization in the ATLAS experiment

Miha Muškinja,John Derek Chapman,Heather Gray,C Doglioni,P Jackson,L Silvestris,W Kamleh,G.A Stewart,D Kim

doi:10.1051/epjconf/202024502036

Miha Muškinja, John Derek Chapman + Show 7 more

Open Access

https://doi.org/10.1051/epjconf/202024502036

Copy DOI

Abstract

Software improvements in the ATLAS Geant4-based simulation are critical to keep up with evolving hardware and increasing luminosity. Geant4 simulation currently accounts for about 50% of CPU consumption in ATLAS and it is expected to remain the leading CPU load during Run 4 (HL-LHC upgrade) with an approximately 25% share in the most optimistic computing model. The ATLAS experiment recently developed two algorithms for optimizing Geant4 performance: Neutron Russian Roulette (NRR) and range cuts for electromagnetic processes. The NRR randomly terminates a fraction of low energy neutrons in the simulation and weights energy deposits of the remaining neutrons to maintain physics performance. Low energy neutrons typically undergo many interactions with the detector material and their path becomes uncorrelated with the point of origin. Therefore, the response of neutrons can be efficiently estimated only with a subset of neutrons. Range cuts for electromagnetic processes exploit a built-in feature of Geant4 and terminate low energy electrons that originate from physics processes including conversions, the photoelectric effect, and Compton scattering. Both algorithms were tuned to maintain physics performance in ATLAS and together they bring about a 20% speed-up of the ATLAS Geant4 simulation. Additional ideas for improvements, currently under investigation, will also be discussed in this paper. Lastly, this paper presents how the ATLAS experiment utilizes software packages such as Intel’s VTune to identify and resolve hot-spots in simulation.

Highlights

Detector simulation is an essential tool for data analysis and the interpretation of physics measurements in High Energy Physics (HEP) experiments such as ATLAS
Physics processes under study are generated with dedicated Monte Carlo (MC) event generator software packages which pseudo-randomly provide final state particles in the detector
In this work we studied the effect of extending the range cuts to compton, photo-electric, and conversion processes by using the energy threshold values already tuned previously for each material-volume pair

Summary

Introduction

Detector simulation is an essential tool for data analysis and the interpretation of physics measurements in High Energy Physics (HEP) experiments such as ATLAS. For the MC production used for Run 2 data analysis, ATLAS spent around 50% of CPU resources on Geant simulation because the majority of background samples were simulated with Geant. Fast simulation techniques are being developed and it is expected that the majority of MC events used in physics analysis will be simulated using these techniques. Even though most of MC events will be simulated by fast simulations, Geant simulation will be the largest CPU consumer with a fraction of about 25%

CPU resource needs

Findings

Summary