The reconstruction of charged particle trajectories is one of the most computationally intensive tasks within current and future filter farms of large High-Energy Physics (HEP) experiments. Due to the increasing number of simultaneous collisions in future high-luminosity colliders, like the HL-LHC and FCC-hh, the challenge of online tracking and event reconstruction becomes even more significant and requires innovative algorithms and appropriate hardware choices for its acceleration. The General Triplet Track Fit is a novel parallelizable track-fitting algorithm that offers a great potential for speed-up by processing triplets of hits independently and allowing to factorize the track reconstruction chain into a detectordependent and independent parts. FPGAs, with their inherent parallelism, power efficiency, and reconfigurability, are becoming increasingly attractive as co-processors for large data centres, such as heterogeneous online farms, to meet the challenges of increasing throughput and computational complexity. A preliminary FPGA implementation of the General Triplet Track Fit has been developed using High-Level Synthesis on AMD FPGAs. Synthesis results indicate that with float precision, a throughput of approximately 107 track fits per second can be achieved, with further room for improvement. The method’s versatility across diverse detector types and its capability to reject fake triplets hold early promise for robust performance in future high-energy physics experiments.
Tastepe et al. (Tue,) studied this question.