Assesment, integration and implementation of computationally efficient models to simulate biological neuronal networks on parallel hardware

The simulation of large-scale biological spiking neural networks (SNN) is computationally onerous. In this work we develop a simulation tool exploiting the computational capabilities of modern graphic processors (GPUs) to speed up simulations of SNNs closely mimicking physiological phenomena. Different models of these phenomena are analyzed, and the best in terms of predictions and compatibility with the SIMD architecture of GPUs are implemented. The performances of the simulator are evaluated.