Granular layEr Simulator: Design and Multi-GPU Simulation of the Cerebellar Granular Layer

Abstract

In modern computational modeling, neuroscientists need to reproduce long-lasting activity of large-scale networks, where neurons are described by highly complex mathematical models. These aspects strongly increase the computational load of the simulations, which can be efficiently performed by exploiting parallel systems to reduce the processing times. Graphics Processing Unit (GPU) devices meet this need providing on desktop High Performance Computing. In this work, authors describe a novel Granular layEr Simulator development implemented on a multi-GPU system capable of reconstructing the cerebellar granular layer in a 3D space and reproducing its neuronal activity. The reconstruction is characterized by a high level of novelty and realism considering axonal/dendritic field geometries, oriented in the 3D space, and following convergence/divergence rates provided in literature. Neurons are modeled using Hodgkin and Huxley representations. The network is validated by reproducing typical behaviors which are well-documented in the literature, such as the center-surround organization. The reconstruction of a network, whose volume is 600 × 150 × 1,200 μm3 with 432,000 granules, 972 Golgi cells, 32,399 glomeruli, and 4,051 mossy fibers, takes 235 s on an Intel i9 processor. The 10 s activity reproduction takes only 4.34 and 3.37 h exploiting a single and multi-GPU desktop system (with one or two NVIDIA RTX 2080 GPU, respectively). Moreover, the code takes only 3.52 and 2.44 h if run on one or two NVIDIA V100 GPU, respectively. The relevant speedups reached (up to ~38× in the single-GPU version, and ~55× in the multi-GPU) clearly demonstrate that the GPU technology is highly suitable for realistic large network simulations.

Cytoarchitectonic Maps of the Human Metathalamus in 3D Space

“…Such a relationship is not only relevant to support MR measurements with the cellular architecture, but also to develop better and more realistic human brain models. The incorporation of cytoarchitectonic parameters has recently led to more biologically valid models of the macaque visual system including cortical areas of different architectural types ( Schmidt et al, 2018 ), as well as models of the human cerebellar granular layer ( Florimbi et al, 2021 ). However, such models usually lack quantitative metathalamic input parameters, forcing them to be estimated indirectly based on other network parameters ( Schmidt et al, 2018 ).…”

Towards the Simulation of a Realistic Large-Scale Spiking Network on a Desktop Multi-GPU System

“…Recently, Granular layEr Simulator (GES) was proposed [ 1 ], which targets multi-GPU systems. This simulator has the advantage of placing the neurons in a three-dimensional space and connecting the elements following biologically realistic rules.…”

“…As a result, the obtained network features a high level of realism since it reproduces the topology of a real granular layer network. However, the work in [ 1 ] describes the single cells activity only through the Hodgkin–Huxley (HH) model, which features a high computational complexity [ 4 ]. This is because the HH model describes the cellular activity in a very realistic way and thus requires a set of differential equations and a high number of parameters.…”

“…Thus, this simulator does not allow to reproduce huge networks on desktop systems, since the GPU memory is limited. In this work, we adopt the same network reconstruction method described in [ 1 ] and simulate the cellular activity with LIF models, which are characterized by a differential equation and feature less parameters compared to the HH description. In this way, we extend the functionality of the original GES simulator, allowing the user to choose the preferred model for network simulation.…”

“…In this way, we extend the functionality of the original GES simulator, allowing the user to choose the preferred model for network simulation. Moreover, since the adopted network is the same as [ 1 ], the computational complexity of parallel performance can be directly compared.…”