Parallel processing of radio signals and detector arrays in CORSIKA 8

TL;DR

This paper discusses recent advances in parallelizing radio signal simulation in CORSIKA 8, including the use of a new multithread scheduler, Gyges, to improve performance and scalability in modeling particle showers.

Contribution

It introduces Gyges, a flexible multithread scheduler, and details its integration into CORSIKA 8 for efficient parallel processing of radio signals.

Findings

Enhanced scalability demonstrated in performance tests

Effective management of worker threads with Gyges

Improved simulation efficiency for radio signals

Abstract

This contribution describes some recent advances in the parallelization of the generation and processing of radio signals emitted by particle showers in CORSIKA 8. CORSIKA 8 is a Monte Carlo simulation framework for modeling ultra-high energy particle cascades in astroparticle physics. The aspects associated with the generation and processing of radio signals in antennas arrays are reviewed, focusing on the key design opportunities and constraints for deployment of multiple threads on such calculations. The audience is also introduced to Gyges, a lightweight, header-only and flexible multithread self-adaptive scheduler written compliant with C++17 and C++20, which is used to distribute and manage the worker computer threads during the parallel calculations. Finally, performance and scalability measurements are provided and the integration into CORSIKA 8 is commented.