Münster 2017 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
HK: Fachverband Physik der Hadronen und Kerne
HK 12: Heavy Ion Collisions and QCD Phases III
HK 12.5: Vortrag
Dienstag, 28. März 2017, 12:15–12:30, F 1
Performance of 4-Dimensional Cellular Automaton Track Finder in CBM — •Valentina Akishina1,3,4 and Ivan Kisel1,2,3 for the CBM collaboration — 1Goethe- Universitat Frankfurt am Main — 2Frankfurt Institute for Advanced Studies — 3GSI Helmholtzzentrum fur Schwerionenforschung GmbH — 4Joint Institute for Nuclear Research
The CBM experiment will focus on the measurement of rare probes at interaction rates up to 10 MHz with data flow of up to 1 TB/s. In this case resolving different collisions, which may overlap in time, is a non-trivial task. Event building requires full online event reconstruction and selection taking into account not only space coordinates, but also time measurements, so-called 4D reconstruction. The algorithms must be fast, precise and suitable for online data processing in order to use the full potential of modern many-core computer architectures.
For the most time-consuming part of the reconstruction procedure the Cellular Automaton track finder is used. The event-based CA track finder was adapted for time-slice-based 4D track reconstruction, which is a requirement in case of CBM. The 4D CA track finder is both vectorized and parallelized. The algorithm shows strong scalability on many-core systems. The speed-up factor of 10.1 was achieved on a CPU with 10 hyper-threaded cores. The algorithm performance is compared with event-based approach event-wise with the help of ideal event-builder. The 4D CA track finder is able to reproduce the efficiency and the speed performance of the event-based CA track finder. The algorithm was included into the CBMROOT framework.