ΕxaNeSt INFN Data Traffic produced by proprietary DPSNN application

DPSNN-STDP is a natively distributed mini-application benchmark representative of plastic spiking neural network simulators. Processes describe synapses in input to cluster of neurons with an irregular interconnection topology, with complex inter-process traffic patterns broadly varying in time and per process. This dataset is comprised of textual data set describing all the details of the inter-processor communication for five different neural network configurations. Each file contains a 3D numerical matrix describing, for each millisecond time-step of simulated neural activity (and for a total duration of 3 seconds of simulation) the size of the payload to be exchanged between all pairs of processes.