Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 41: Trigger+DAQ 1

T 41.6: Talk

Tuesday, March 5, 2024, 17:15–17:30, Geb. 30.23: 3/1

Comparison of a Linearized Track Fitting Algorithm, Implemented on GPU and FPGA — •Joachim Zinßer, Sebastian Dittmeier, and André Schoening — Physikalisches Institut, Heidelberg, Germany

For the Event Filter System of the ATLAS experiment at the High-Luminosity LHC, studies on accelerating the online track reconstruction for the Inner Tracker (ITk) are conducted, which possibly lead to a heterogeneous system using GPUs or FPGAs. This study focusses on the implementation of one block of a linearized track fitting pipeline that can be used for quick fake track rejection and track parameter estimation. This algorithm can be implemented in a highly parallel way and is, therefore, suited for hardware acceleration. The algorithm is implemented on a Intel Stratix 10 FPGA and a NVIDIA A6000 GPU are used. The physics performance of the two implementations, are compared to the results from the fast emulation as presented in the ATL-DAQ-PUB-2023-001 document, using the same data. Furthermore, the computing performance are compared with each other.

Keywords: ATLAS High-Lumi Upgrade; Event Filter System; Parallel Computing; Linearized Track Fitting