Bluehive and OSB interactions¶

Discussion of the options for running models from OSB (and any other models expressed in LEMS/NeuroML2) on the Bluehive hardware being developed in Cambridge.

Some of these points may be applicable to other FPGA based simulator systems.

Overview of Bluehive system¶

Information on the current status of the Bluehive system can be found in presentations by Steve Marsh and Paul Fox at the 2013 OSB kickoff meeting

More information…

Potential OSB models to test on Bluehive hardware¶

Some of the potential models on OSB which could be tried out on Bluehive include:

• Sparsely connected spiking neuron network - Brunel 2000
• Self Sustained Network Activity - Destexhe 2009 (Note these are both PyNN based networks, which could potentiall be scaled up by changing the population size parameters)
• Granule Cell Layer - Maex and De Schutter 1998 (This is a neuroConstruct based project containing a network of single compartment conductance based neurons. The Large 3D network simulation configuration could be used for testing on Bluehive, but would require more testing of the ChannelML > NeuroML2> LEMS -> Other formats pathway)

Note that a Google Summer of Code project for the INCF will start soon when a student will be specifically looking at converting large scale cortical models to NeuroML/PyNN and putting them on OSB: http://www.incf.org/gsoc/2013, http://www.google-melange.com/gsoc/project/google/gsoc2013/vitorchaud/14001.

These networks will be exactly the type needed for testing with Bluehive and comparing to other simulators.

More…

Libraries which can be used for code generation, testing, etc.¶

PyLEMS
Extended org.neuroml.export package which exports VFGEN format, to generate C/C++…

Other simulator systems based on FPGA¶

Please add comments if you know anything about the state of these systems (but be nice, it’s public!).

A Large-Scale Spiking Neural Network Accelerator for FPGA Systems Systems
Kit Cheung, Simon R. Schultz, Wayne Luk, Imperial College London

A Biologically Plausible Real-time Spiking Neuron Simulation Environment Based on a Multiple-FPGA Platform
Shufan Yang, TM McGinnity, University of Ulster

An FPGA-based approach to high-speed simulation of conductance-based neuron models
E. L. Graas, E. A. Brown, Robert H. Lee, Emory University/Georgia Institute of Technology