Cellular automaton based track reconstruction in the STAR TPC

STAR is an active collider heavy-ion experiment at RHIC/BNL (Upton, USA). The main tracking detector of the experiment is a Time Projection Chamber (TPC). For future collision rates of up to 1 kHz STAR experiment requires a fast track reconstruction procedure, which can deal with high track densities (up to 10-20 thousands tracks). The Cellular Automaton (CA) algorithm is based on local reconstruction and therefore is robust, fast and easily parallelized, that makes it perfectly suitable for the task under consideration. The algorithm has been implemented by adaptation of the Alice HLT TPC CA track finder for the STAR TPC detector. The efficiency, speed and robustness have been increased. The algorithm is highly parallelized on both data (uses SIMD instruction set) and task (uses Threading Building Blocks technology) levels. The memory usage in the algorithm is optimized as well. Tests with Au-Au at 200 AGeV real data events have been performed. It has been shown that event reconstruction procedure with the CA based track finder demonstrates better track reconstruction efficiency by 9% with respect to the previous scalar version based on track following with Kalman Filter. CA based track finder takes only 10% of event reconstruction procedure.