Dataset for paper "Larger GPU-accelerated brain simulations with procedural connectivity"

<b>Dataset for paper published in Nat Comput Sci Feb 2021</b><b><br></b>Dataset contains raw spiking data from full-scale multi-area model simulation run using GeNN 4.3.3. Each tar.gz archive contains the configuration files for each simulation and, in the recording directory, binary numpy files contains the spike trains from each population.<br>Archives with filenames starting with <b>82d3c0816b0ad1c07ea27e61eb981f7a </b>contain spike data from three 10.5 second "ground state" simulations of the model's "ground state" (chi=1.0)<br>Archives with filenames starting with <b>b03fdaa1fd47a0e4a10483bc3901f1e5 </b>contain spike data from three 100.5 second "ground state" simulations of the model's "resting state" (chi=1.9)<br><br><b>Abstract</b><br>"Simulations 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. Extending our GeNN software with procedural connectivity and a second technical innovation for GPU code generation, we can simulate a recent model of the Macaque visual cortex with 4.13⁶ neurons and 24.2⁹synapses on a single GPU - a significant step forward in making large-scale brain modelling accessible to more researchers."<br>FundingBrains on Board grant number EP/P006094/1<br><br>