Author(s)
|
Schmidt, Mustafa Andre (Bergische Universitaet Wuppertal (DE)) ; Vishwakarma, Akanksha (The University of Edinburgh (GB)) ; Sukharev, Andrei (Budker Institute of Nuclear Physics (RU)) ; Wynne, Benjamin Michael (The University of Edinburgh (GB)) ; Morgan, Benjamin (University of Warwick (GB)) ; Marcon, Caterina (Università degli Studi e INFN Milano (IT)) ; Kim, Dongwon (Stockholm University (SE)) ; Kourlitis, Vangelis (Technische Universitat Munchen (DE)) ; Tcherniaev, Evgueni (University of Pittsburgh (US)) ; Amadio, Guilherme (CERN) ; Apostolakis, John (CERN) ; Chapman, John Derek (University of Cambridge (GB)) ; Bandieramonte, Marilena (University of Pittsburgh (US)) ; Muskinja, Miha (Jozef Stefan Institute (SI)) ; Novak, Mihaly (CERN) ; Lari, Tommaso (Università degli Studi e INFN Milano (IT)) ; Hopkins, Walter (Argonne National Laboratory (US)) |
Abstract
| Several optimization techniques are implemented in the ATLAS Geant4 detector simulation to improve CPU and memory usage. These optimizations include tracking methods for gammas, the usage of static linking, adjustments to electromagnetic range cuts, a Russian Roulette process for reducing simulation steps, and advanced geometry management. The latest performance benchmarks show significant improvements in simulation speed and resource efficiency, crucial for future LHC high-luminosity operations. Consequently, more projects are initiated to improve the ATLAS full simulations even further, mainly concerning improving the CPU time. This paper summarizes the validated and ongoing tasks as well as the obtained performance increase. |