Speakers
Description
Event-processing frameworks are an important part of High Energy Physics (HEP) software, providing the infrastructure for a wide range of domain-specific tasks such as event reconstruction and simulation. They are designed to deliver high throughput and meet the demanding performance requirements of the field. A key component of modern frameworks is the scheduler, which enables efficient parallelization of complex data processing workflows. The Julia programming language, with its focus on high performance and a growing ecosystem for scientific computing, presents an interesting option for implementing such frameworks.
In this work, we focus on the performance of demonstrator project for an event-processing framework developed in Julia and using Dagger.jl, a library for parallel and heterogeneous processing. We specifically evaluate its multithreaded performance and scalability. The discussion will highlight both the advantages and challenges of using Dagger and Julia for this type of HEP application, providing insights into their potential for future event-processing frameworks.
