Multi-level Parallelization in ROOTNew Patterns, Libraries and Utilities

Resumen

TThe Large Hadron Collider (LHC) is generating Physics data at an unprecedented rate that is expected to continue increasing in the future. This situation results in increasing demands in computation and processing power for analyzing the LHC data. At the same time, modern architectures have switched from a sequential processing paradigm to a parallel one. In order to take full advantage of new computer architectures, to improve performance with increasing amounts of data to analyze, and to reduce development time and complexity, ROOT, the official LHC analysis framework, is undertaking a modernization effort to cope with the computational challenges posed by the ambitious physics program of the LHC. We contribute to this modernization effort by implementing patterns, introducing libraries and developing utilities for the parallelization of ROOT at multiple levels, improving the programming model and providing backward compatibility.