Architecture and Performance of the Inner Detector Trigger of the ATLAS detector

The rising instantaneous luminosity of the LHC poses an increasing challenge to the pattern recognition algorithms for track reconstruction at the ATLAS Inner Detector Trigger. The tracking trigger of ATLAS has been so far using two strategies, of similar performance and both fast enough to respect the tight computation time constraint imposed by the online decision. IDScan and SiTrack algorithms are the two main pattern recognition algorithms, respectively based on a histogramming and a combinatorial approaches of hits in the silicon tracking of the Inner Detector. The IDScan algorithm is used to reconstruct tracks in electron and muon trigger signatures. It shows very good performance in late 2011 data, as presented here. In light of the experience from the data taking in 2011, a new software framework, called L2Star, is being developed that will provide a suite of configurable tools, based on modularising the existing code and increasing the re-use of components, to provide the optimal solution in the various trigger signatures at higher luminosities. The architecture of the L2Star platform is presented.