SMuK 2023 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

AKPIK: Arbeitskreis Physik, moderne Informationstechnologie und Künstliche Intelligenz

AKPIK 2: Applications in Particle and Astroparticle Physics

AKPIK 2.2: Vortrag

Dienstag, 21. März 2023, 17:15–17:30, ZEU/0118

Providing GPU resources in a HEP analysis environment — Johannes Erdmann¹, Benjamin Fischer¹, Thomas Kreß², Dennis Noll¹, Andreas Nowack², and •Roman Suveyzdis¹ — ¹III. Physikalisches Institut A, RWTH Aachen University — ²III. Physikalisches Institut B, RWTH Aachen University

Graphics Processing Units (GPUs) have become a key computing resource for advanced physics analyses for example for training and evaluating machine learning models. The local research group’s computing cluster is still often limited in GPU resources. Therefore, an efficient and fair use is crucial. HTCondor, a powerful batch job software system that is often used in the HEP physics community, offers the possibility to manage GPU resources. The basic installation of HTCondor only distributes entire GPUs on a per job basis, which leaves some potential resources unused. In this talk, we will present how HTCondor can be configured to cope with the users’ need to use GPUs both interactively and in batch job mode. We will report on the first experiences with our setup.