Speaker
Description
ALICE has upgraded its computing infrastructure and software to accommodate the increased data rates and enhanced detector performance in Run 3. This upgrade has also resulted in a significant reduction in energy consumption and CO₂ emissions, achieved through a combination of optimized resource utilization and the adoption of energy-efficient technologies. Key strategies include maximizing GPU utilization, parallelizing and vectorizing CPU code, and organizing data analysis workflows into trains, along with the use of low-PUE data centers for analysis tuning and validation. The reuse of generated events in Monte Carlo simulations and the integration of ARM architecture support have further contributed to energy savings. Moreover, enhancements to the EPN infrastructure, including the implementation of energy-efficient cooling systems, have led to substantial cost reductions. These strategies will be presented and discussed in detail during the talk.
