KIT | KIT-Bibliothek | Impressum | Datenschutz

KINGFISHER: A Framework for Fast Machine Learning Inference for Autonomous Accelerator Systems

Scomparin, Luca ORCID iD icon 1; Blomley, Edmund ORCID iD icon 2; Bründermann, Erik ORCID iD icon 2; Caselle, Michele 1; Dritschler, Timo ORCID iD icon 1; Kopmann, Andreas ORCID iD icon 1; Mochihashi, Akira 2; Müller, Anke-Susanne ORCID iD icon 2; Santamaria Garcia, Andrea ORCID iD icon 2; Schreiber, Patrick ORCID iD icon 2; Steinmann, Johannes L. ORCID iD icon 2; Weber, Marc 1; Boltz, T.
1 Institut für Prozessdatenverarbeitung und Elektronik (IPE), Karlsruher Institut für Technologie (KIT)
2 Institut für Beschleunigerphysik und Technologie (IBPT), Karlsruher Institut für Technologie (KIT)

Abstract (englisch):

Mod­ern par­ti­cle ac­cel­er­a­tor fa­cil­i­ties allow new and ex­cit­ing beam prop­er­ties and op­er­a­tion modes. Tra­di­tional real-time con­trol sys­tems, al­beit pow­er­ful, have band­width and la­tency con­straints that limit the range of op­er­at­ing con­di­tions cur­rently made avail­able to users. The ca­pa­bil­ity of Re­in­force­ment Learn­ing to per­form self-learn­ing con­trol poli­cies by in­ter­act­ing with the ac­cel­er­a­tor is in­trigu­ing. The ex­treme dy­namic con­di­tions re­quire fast real-time feed­back through­out the whole con­trol loop from the di­ag­nos­tic, with novel and in­tel­li­gent de­tec­tor sys­tems, all the way to the in­ter­ac­tion with the ac­cel­er­a­tor com­po­nents. In this con­tri­bu­tion, the novel KING­FISHER frame­work based on the mod­ern Xil­inx Ver­sal de­vices will be pre­sented. Ver­sal com­bines sev­eral com­pu­ta­tional en­gines, specif­i­cally com­bin­ing pow­er­ful FPGA logic with pro­gram­ma­ble AI En­gines in a sin­gle de­vice. Fur­ther­more, this sys­tem can be na­tively in­te­grated with the fastest beam di­ag­nos­tic tools al­ready avail­able, i.e. KAP­TURE and KA­LYPSO.

DOI: 10.18429/JACoW-IBIC2022-MOP42
Zitationen: 1
Zugehörige Institution(en) am KIT Institut für Beschleunigerphysik und Technologie (IBPT)
Institut für Prozessdatenverarbeitung und Elektronik (IPE)
Publikationstyp Proceedingsbeitrag
Publikationsdatum 09.10.2022
Sprache Englisch
Identifikator ISBN: 978-3-95450-241-7
ISSN: 2673-5350
KITopen-ID: 1000154412
HGF-Programm 54.12.03 (POF IV, LK 01) Science Systems
Weitere HGF-Programme 54.11.11 (POF IV, LK 01) Accelerator Operation, Research and Development
Erschienen in Data Acquisition and Processing Platforms - International Beam Instrumentation Conference (11th), Kraków, Poland, 11-15 September 2022
Veranstaltung 11th International Beam Instrumentation Conference (IBIC 2022), Krakau, Polen, 11.09.2022 – 15.09.2022
Verlag JACoW Publishing
Seiten 151-155
Externe Relationen Konferenz
Nachgewiesen in Scopus
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page