In preparation for Run 3 at the Large Hadron Collider, the Monte Carlo Detector Simulation performed with Geant4 (Full Simulation) within the ATLAS experiment has undergone significant improvements to enhance its computational performance and overall efficiency. These proceedings offer a comprehensive overview of the optimisations implemented in the ATLAS Full Simulation for Run 3. Notable developments include the application of EM range cuts, the implementation of Neutron and Photon Russian roulette and the recent development of the Woodcock tracking in the Electromagnetic Endcap Calorimeter, a critical subdetector that for its peculiar structure represents a significant consumer of CPU resources in the ATLAS simulation.
Beyond optimisations targeting specific detector components, the improvements include the tuning of simulation parameters (including the magnetic field), smarter and more efficient geometry descriptions, the implementation of new Geant4 core features and improvements that target the way Geant4 is linked and used within the ATLAS simulation framework. These enhancements collectively resulted in a remarkable achievement, with a speed-up in CPU time of a factor of 2 compared to the baseline simulation configuration used in Run 2. In addition to showcasing the Run 3 ATLAS simulation, this contribution provides an overview of forthcoming optimisations, emphasising both immediate and longer-term enhancements.