Neuron Models in FPGA Hardware - A Route from High Level Descriptions to Hardware Implementations

Abstract

This paper presents the LEMS2HDL toolsuite which converts Low Entropy Model Specification (LEMS) neuron/neural network models to synthesisable Hardware Description Language (HDL) hardware descriptions. The LEMS2HDL process will provide a route for the neuroscience community to perform accelerated Field-Programmable Gate Array (FPGA) hardware implementations of the growing library of LEMS neuron/neural network models. The paper describes the LEMS to HDL conversion process and references the previously reported vicilogic platform. The paper compares the resulting FPGA hardware simulation of three LEMS neuron models with the LEMS model simulation.