IN-IDE-2017-001

The construction of large scale high energy physics experiments present many
exciting challenges in various fields. One crucial task the development of technology
has to handle is the immense amount of data generated by detectors in accelerator
experiments. This paper deals with the upcoming PANDA experiment currently
in construction in Darmstadt, Germany, which is expected to generate data at a
rate of 200 GB/s. The size of the data stream has to be significantly reduced
by a triggering system, which filters the data before storage. Triggering systems
have traditionally been implemented in hardware, by tuning the detectors to only
collect data for short time windows when signals are registered. However, due to
the high technical demands, the PANDA experiment will use an entirely software-
based trigger. The filtering will be performed using algorithms to reconstruct physical
observables from the raw data which are then used to select the interesting data
segments for storage.
This paper describes the implementation of a prototype for real-time software
triggering for a subdetector of the PANDA experiment. The prototype is highly
scalable and implemented with the use of the MPI and OpenMP frameworks. In addition, new algorithms for early event reconstruction based on raw data are
presented.