Hardware-Portable Data Analysis Building Blocks for the High Luminosity LHC Era

Upcoming High Luminosity experiments at the Large Hadron Collider (HL-LHC) will deliver more than a magnitude more data than current experiments. This big data necessitates new high-performance software solutions to analyze the data and ensure the portability of the software to different hardware devices, such as various kinds of accelerators (GPUs).
In the present project, we attempt to explore and investigate hardware-portable ecosystems such as ALPAKA and KOKKOS for HL-LHC data analyses. The outcome will allow us to learn which ecosystem is best suited for expected analyses. Furthermore, it will elucidate which related data access patterns exist within the respective data handling routines and which will benefit from accelerators. A subset of the identified patterns shall be implemented using one of these hardware-portable ecosystems and, as building blocks, be provided to the community - a starting point for sustainable, hardware-portable HPC-aware software building blocks in HL-LHC data analysis.
Close interactions with the communities on high-performance computing and LHC computing will be sought. Code evaluations will be carried out on the large-scale computing facilities at Deutsches Elektronen-Synchrotron DESY.