Speaker
Description
Event processing frameworks are important software components of High Energy Physics (HEP) experiments, playing a critical role in building applications for HEP-specific workflows such as trigger or event reconstruction. A key aspect of these frameworks is their ability to efficiently orchestrate the parallel processing of algorithms for multiple events simultaneously. As heterogeneous setups, including GPUs and other accelerators, become increasingly accessible, there is a growing need to incorporate these resources into the frameworks effectively.
Our demonstrator project investigates the viability of developing an event-processing framework that utilizes heterogeneous resources in the Julia language. For the first demonstrator, we chose Dagger.jl as a library supporting parallel and heterogeneous computing. We will present the general assumptions and requirements for a framework, address the current state of the demonstrator project and its methodology, as well as discuss our experience with Julia's ecosystem and Dagger.jl, highlighting the strengths and challenges we encountered.
