Online track and vertex reconstruction on GPUs for the Mu3e experiment

The Mu3e experiment searches for the lepton flavour violating decay μ → eee, aiming at a branching ratio sensitivity better than 10^-16.To reach this sensitivity, muon rates above 10⁹ μ/s are required. A high precision silicon tracking detector combined with excellent timing resolution from scintillating fibers and tiles will measure the momenta, vertices and timing of the decay products of muons stopped in the target to suppress background. The trigger-less readout system will deliver about 100 GB/s of zero-suppressed data. A network of optical links and switching FPGAs sends the complete detector data for a time slice to one node of the filter farm. An FPGA inside the filter farm PC transfers the event data to the GPU via PCIe direct memory access. The GPU finds and fits tracks using a 3D tracking algorithm for multiple scattering dominated resolution. In a second step, a three track vertex fit is performed, allowing for a reduction of the output data rate to below 100 MB/s by removing combinatorial background. The talk discusses the data flow from the FPGA to the GPU as well as the implementation and performance of the track and vertex fits on the GPU.