Gast ::
| Hauptseite > Publikationsdatenbank > Supercomputers ready for use as discovery machines for neuroscience |
| Hauptseite > Publikationsdatenbank > Supercomputers ready for use as discovery machines for neuroscience |
| Contribution to a conference proceedings | FZJ-2013-03827 |
; ; ; ; ; ; ; ; ; ;
2013
Abstract: NEST is a widely used tool to simulate biological spiking neural networks [1]. The simulator is subject to continuous development, which is driven by the requirements of the current neuroscientific questions. At present, a major part of the software development focuses on the improvement of the simulator's fundamental data structures in order to enable brain-scale simulations on supercomputers such as the Blue Gene system in Jülich and the K computer in Kobe. Based on our memory-usage model [2], we redesigned the neuronal and the connection infrastructure of NEST such that networks of 10^8 neurons and 10^12 synapses can be simulated on the K computer [3]. These improvements reduce the memory footprint without compromising on the simulator's general usability and user interface. Here, we describe the recent technological advances which enable NEST to achieve high performance and good scaling of network setup and simulation on the K computer and on the Blue Gene system. We demonstrate that the usability of these machines for network simulations has become comparable to running simulations on a single PC.
|
The record appears in these collections: |
