Description
Track reconstruction algorithms are computationally intensive due to their combinatorial nature and pose a great challenge for HL-LHC era. The estimated compute time will not fit the budget unless the code becomes more efficient and highly parallel. Exploring heterogeneous architectures is the at core of this change and current R&D efforts (mostly focused on CUDA) show promising results, but the goal hasn't been reached yet. The talk introduces the problem(s), some notable published results and connects this to my PhD research topic.
Speaker
Georgiana Mania
(DESY)