DYNAP-SE1 introduction#

A DYNAP-SE1 board has 4 chips and each chip contains 4 cores featuring 256 silicon neurons. We call the state of the 4 chips a configuration. Each silicon neuron has 64 synapses to receive the incoming connections of its presynaptic neurons. Each core holds a parameter group to store the neuronal and synaptic parameters for its 256 neurons and their incoming synapses. The neurons in the same core share the same parameter values. There are 4 types of synapses: AMPA, NMDA, GABA_A, and GABA_B. 1024 spike generators are available implemented on the FPGA. Each of the spike generator can be used as a Poisson spike generator or a normal FPGA spike generator.

To use DYNAP-SE1, we recommend a utilities library based on Samna, developed primarily by the NCS group of INI. This Python repository adds more features to Samna for DYNAP-SE1, which wraps around the plain Samna classes and functions to provide more user-friendly and convenient Python APIs for constructing networks, configuring hardware parameters, monitoring neurons activity, implementing learning algorithms, etc.