Dynamic parallelism for synaptic updating in GPU-accelerated spiking neural network simulations

Kasap, Bahadir; Opstal, A. John Van

doi:10.1016/j.neucom.2018.04.007

Cited by 35 publications

(19 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The motor map is represented as a rectangular grid of neurons with a Mexican hat-type pattern of lateral interactions. The neural activities were simulated by custom code utilizing dynamic parallelism to accelerate spike propagation on a GPU [41]. The code was developed and tested on a Tesla K40 with CUDA Toolkit 7.0, Linux Ubuntu 16.04 LTS (repository under https://bitbucket.org/bkasap/sc_microstimulation).…”

Section: Methodsmentioning

confidence: 99%

Microstimulation in a spiking neural network model of the midbrain superior colliculus

Kasap

Opstal

2019

PLoS Comput Biol

Self Cite

View full text Add to dashboard Cite

The midbrain superior colliculus (SC) generates a rapid saccadic eye movement to a sensory stimulus by recruiting a population of cells in its topographically organized motor map. Supra-threshold electrical microstimulation in the SC reveals that the site of stimulation produces a normometric saccade vector with little effect of the stimulation parameters. Moreover, electrically evoked saccades (E-saccades) have kinematic properties that strongly resemble natural, visual-evoked saccades (V-saccades). These findings support models in which the saccade vector is determined by a center-of-gravity computation of activated neurons, while its trajectory and kinematics arise from downstream feedback circuits in the brainstem. Recent single-unit recordings, however, have indicated that the SC population also specifies instantaneous kinematics. These results support an alternative model, in which the desired saccade trajectory, including its kinematics, follows from instantaneous summation of movement effects of all SC spike trains. But how to reconcile this model with microstimulation results? Although it is thought that microstimulation activates a large population of SC neurons, the mechanism through which it arises is unknown. We developed a spiking neural network model of the SC, in which microstimulation directly activates a relatively small set of neurons around the electrode tip, which subsequently sets up a large population response through lateral synaptic interactions. We show that through this mechanism the population drives an E-saccade with near-normal kinematics that are largely independent of the stimulation parameters. Only at very low stimulus intensities the network recruits a population with low firing rates, resulting in abnormally slow saccades.

show abstract

Section: Methodsmentioning

confidence: 99%

Microstimulation in a spiking neural network model of the midbrain superior colliculus

Kasap

Opstal

2019

PLoS Comput Biol

Self Cite

View full text Add to dashboard Cite

show abstract

“…The effect of ASG on simulation time is explored in the results section below, where it is compared to a basic synapse update algorithm and a recently proposed GPU specific optimisation, referred to here as Dynamic Synapse Parallelism (DSP) (Kasap and van Opstal, 2018).…”

Section: Active Synapse Groupingmentioning

confidence: 99%

“…This is called Dynamic Parallelism. Dynamic Parallelism was identified by Kasap and van Opstal (2018) as a potential speedup when compared to a limited set of other synaptic update methods.…”

Section: The Spike Simulator and Optimisationsmentioning

confidence: 99%

See 1 more Smart Citation

Spike: A GPU Optimised Spiking Neural Network Simulator

Ahmad

Isbister

Smithe

et al. 2018

Preprint

View full text Add to dashboard Cite

Spiking Neural Network (SNN) simulations require internal variables -such as the membrane voltages of individual neurons and their synaptic inputs -to be updated on a sub-millisecond resolution. As a result, a single second of simulation time requires many thousands of update calculations per neuron. Furthermore, increases in the scale of SNN models have, accordingly, led to manyfold increases in the runtime of SNN simulations. Existing solutions to this problem of scale include high performance CPU based simulators capable of multithreaded execution ("CPU parallelism"). More recent GPU based simulators have emerged, which aim to utilise GPU parallelism for SNN execution. We have identified several key speedups, which give GPU based simulators up to an order of magnitude performance increase over CPU based simulators on several benchmarks. We present the Spike simulator with three key optimisations: timestep grouping, active synapse grouping, and delay insensitivity. Combined, these optimisations massively increase the speed of executing a SNN simulation and produce a simulator which is, on a single machine, faster than currently available simulators.

show abstract

“…The motor map is represented as a rectangular grid of neurons with 221 a Mexican hat-type pattern of lateral interactions. The neural activities were simulated by custom code 222 utilizing dynamic parallelism to accelerate spike propagation on a GPU (Kasap and Van Opstal, 2018). 223…”

Section: The Adex Neuron Model 219mentioning

confidence: 99%

Microstimulation in a spiking neural network model of the midbrain superior colliculus

Opstal

Kasap

2018

Preprint

Self Cite

View full text Add to dashboard Cite

13The midbrain superior colliculus (SC) generates a rapid saccadic eye movement to a sensory stimulus 14 by recruiting a population of cells in its topographically organized motor map. Supra-threshold 15 electrical microstimulation in the SC reveals that the site of stimulation produces a normometric 16 saccade vector with little effect of the stimulation parameters. Moreover, electrically evoked saccades 17 (E-saccades) have kinematic properties that strongly resemble natural, visual-evoked saccades (V-18 saccades). These findings support models in which the saccade vector is determined by a center-of-19 gravity computation of activated neurons, while its trajectory and kinematics arise from downstream 20 feedback circuits in the brainstem. Recent single-unit recordings, however, have indicated that the SC 21 population also specifies instantaneous kinematics. These results support an alternative model, in 22 which the desired saccade trajectory, including its kinematics, follows from instantaneous summation 23 Kasap and Van Opstal: Microstimulation in a spiking neural network model 2 of movement effects of all SC spike trains. But how to reconcile this model with microstimulation 24 results? Although it is thought that microstimulation activates a large population of SC neurons, the 25 mechanism through which it arises is unknown. We developed a spiking neural network model of the 26 SC, in which microstimulation directly activates a relatively small set of neurons around the electrode 27 tip, which subsequently sets up a large population response through lateral synaptic interactions. We 28 show that through this mechanism the population drives an E-saccade with near-normal kinematics 29 that are largely independent of the stimulation parameters. Only at very low stimulus intensities the 30 network recruits a population with low firing rates, resulting in abnormally slow saccades. 31 32 33 34 35 36 37 Author Summary 38 39 The midbrain Superior Colliculus (SC) contains a topographically organized map for rapid goal-40 directed gaze shifts, in which the location of the active population determines size and direction of the 41 eye-movement vector, and the neural firing rates specify the eye-movement kinematics. Electrical 42 microstimulation in this map produces eye movements that correspond to the site of stimulation with 43 normal kinematics. We here explain how intrinsic lateral interactions within the SC network of spiking 44 neurons sets up the population activity profile in response to local microstimulation to explain these 45 results. 46 47 48

show abstract

Dynamic parallelism for synaptic updating in GPU-accelerated spiking neural network simulations

Cited by 35 publications

References 22 publications

Microstimulation in a spiking neural network model of the midbrain superior colliculus

Microstimulation in a spiking neural network model of the midbrain superior colliculus

Spike: A GPU Optimised Spiking Neural Network Simulator

Microstimulation in a spiking neural network model of the midbrain superior colliculus

Contact Info

Product

Resources

About