A Spiking Neural Simulator Integrating Event-Driven and Time-Driven Computation Schemes Using Parallel CPU-GPU Co-Processing: A Case Study

Abstract

Time-driven simulation methods in traditional CPU architectures perform well and precisely when simulating small-scale spiking neural networks. Nevertheless, they still have drawbacks when simulating large-scale systems. Conversely, event-driven simulation methods in CPUs and time-driven simulation methods in graphic processing units (GPUs) can outperform CPU time-driven methods under certain conditions. With this performance improvement in mind, we have developed an event-and-time-driven spiking neural network simulator suitable for a hybrid CPU-GPU platform. Our neural simulator is able to efficiently simulate bio-inspired spiking neural networks consisting of different neural models, which can be distributed heterogeneously in both small layers and large layers or subsystems. For the sake of efficiency, the low-activity parts of the neural network can be simulated in CPU using event-driven methods while the high-activity subsystems can be simulated in either CPU (a few neurons) or GPU (thousands or millions of neurons) using time-driven methods. In this brief, we have undertaken a comparative study of these different simulation methods. For benchmarking the different simulation methods and platforms, we have used a cerebellar-inspired neural-network model consisting of a very dense granular layer and a Purkinje layer with a smaller number of cells (according to biological ratios). Thus, this cerebellar-like network includes a dense diverging neural layer (increasing the dimensionality of its internal representation and sparse coding) and a converging neural layer (integration) similar to many other biologically inspired and also artificial neural networks.

Model-Driven Analysis of Eyeblink Classical Conditioning Reveals the Underlying Structure of Cerebellar Plasticity and Neuronal Activity

“…1 right panel) was composed of 100 Mossy Fibers (MFs), 2000 Granule cells (GRs), 12 Inferior Olive cells (IOs), 12 Purkinje Cells (PCs) and 6 Deep Cerebellar Nuclei cells (DCNs). All the neurons were modeled as leaky integrate-and-fire neurons because they required only a few state variables to be implemented [34], [35], [36]. The MFs received the CS and were randomly connected with the granular layer; each GR received 4 random excitatory synapses from the MFs.…”

“…5A). MFs had a random activity with an average firing rate of 39 [range [32][33][34][35][36][37][38][39][40][41][42][43][44] Hz. IOs were active only during the US (i.e.…”

“…In this way, it was possible to simulate different parameter sets in parallel on the four CPU cores, but this limited the simulation performances. To further reduce the simulation time, maintaining the network dimension, it could be possible to run a single simulation in parallel on multiple CPU cores or using a GPU-approach, which speeds up the simulation [15], [34].…”

Data-Driven Compartmental Modeling Method for Harmonic Analysis—A Study of the Electric Arc Furnace

“…where I 1 denotes the fundamental current. The non-linear function, F h , leads to the heavy computational cost of harmonic power-flow calculation, due to the iterative solution of the system equations and non-linear coupling model [20,21]. Therefore, a non-linear coupling model is approximated to a linear coupling model, as shown in Equation (3).…”

“…In light of the potential benefits of artificial intelligence, e.g., a strong ability for nonlinear mapping, the neural-network-based method was employed to establish the multi-mode EAF harmonic model [18][19][20]. Nevertheless, the non-linear multi-mode EAF harmonic model established by the neural-network-based method is not suitable for the harmonic power-flow calculation because it results in a heavy computational cost in the harmonic power-flow calculation [21,22].…”