procedural.pdf (1.97 MB)
Larger GPU-accelerated brain simulations with procedural connectivity
journal contribution
posted on 2023-06-09, 22:49 authored by James KnightJames Knight, Thomas NowotnyThomas NowotnySimulations are an important tool for investigating brain function but large models are needed to faithfully reproduce the statistics and dynamics of brain activity. Simulating large spiking neural network models has, until now, needed so much memory for storing synaptic connections that it required high performance computer systems. Here, we present an alternative simulation method we call `procedural connectivity' where connectivity and synaptic weights are generated `on the fly' instead of stored and retrieved from memory. This method is particularly well-suited for use on Graphical Processing Units (GPUs) - which are a common fixture in many workstations. Using procedural connectivity and an additional GPU code generation optimisation, we can simulate a recent model of the Macaque visual cortex with 4.13 million neurons and 24.2 billion synapses on a single GPU - a significant step forward in making large-scale brain modelling accessible to more researchers.
Funding
Human Brain Project Specific Grant Agreement 2 — HBP SGA2; G2410; EUROPEAN UNION; 785907
Brains on Board: Neuromorphic Control of Flying Robots; G1980; EPSRC-ENGINEERING & PHYSICAL SCIENCES RESEARCH COUNCIL; EP/P006094/1
HBP SGA3 - Human Brain Project Specific Grant Agreement 3; G2945; EUROPEAN UNION; 945539
History
Publication status
- Published
File Version
- Accepted version
Journal
Nature Computational ScienceISSN
2662-8457Publisher
Nature Publishing GroupExternal DOI
Volume
1Page range
136-142Department affiliated with
- Informatics Publications
Research groups affiliated with
- Centre for Computational Neuroscience and Robotics Publications
Full text available
- Yes
Peer reviewed?
- Yes
Legacy Posted Date
2021-01-19First Open Access (FOA) Date
2021-08-02First Compliant Deposit (FCD) Date
2021-01-18Usage metrics
Categories
No categories selectedLicence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC