A Real-Time Reconfigurable Multichip Architecture for Large-Scale Biophysically Accurate Neuron Simulation

Abstract

Simulation of brain neurons in real-time using biophysically meaningful models is a prerequisite for comprehensive understanding of how neurons process information and communicate with each other, in effect efficiently complementing in-vivo experiments. State-of-the-art neuron simulators are, however, capable of simulating at most few tens/hundreds of biophysically accurate neurons in real-time due to the exponential growth in the interneuron communication costs with the number of simulated neurons. In this paper, we propose a real-time, reconfigurable, multichip system architecture based on localized communication, which effectively reduces the communication cost to a linear growth. All parts of the system are generated automatically, based on the neuron connectivity scheme. Experimental results indicate that the proposed system architecture allows the capacity of over 3000 to 19 200 (depending on the connectivity scheme) biophysically accurate neurons over multiple chips.

Sharpness recognition based on synergy between bio-inspired nociceptors and tactile mechanoreceptors

“…Digital execution with Field-Programmable Gate Array (FPGA) offers parallel computations and flexibility for algorithm investigation while filling time and performance limitations. FPGAs have broad applications in the neural network simulations 31 and motivate further exploration 32,33 . An approximate circuit technique was used to implement tactile data processing on FPGA for the e-skin applications 34 .…”

flexHH: A Flexible Hardware Library for Hodgkin-Huxley-Based Neural Simulations

“…Although per time step the flexHH is faster than the hardcoded version as described before, having such a stricter time-step size than the BrainFrame kernel [7], forces the flexHH kernel to conduct far greater computations for the same simulated brain time. The implementation of Zjajo et al [24] can support a larger network size under real-time conditions as it would be expected with the reduced connectivity density.…”

“…Here, the flexHH library provides more than twice the performance density for the simple HH case and about 65% higher FLOPS/LUT for the IO implementation compared to the BrainFrame hardcoded design. Compared to the traditional FPGA-based platform of Zjajo et al [24], flexHH provides more than 5× higher performance density when simulating at real-time speeds. Consequently, dataflow programming also favors performance efficiency.…”

Sharpness Recognition Based on Synergy between Bio-inspired Nociceptors and Tactile Mechanoreceptors 

“…Digital execution with Field-Programmable Gate Array (FPGA) offers parallel computations and exibility for algorithm investigation while lling time and performance limitations. FPGAs have broad applications in the neural network simulations 31 and motivate further exploration 32,33 . An approximate circuit technique was used to implement tactile data processing on FPGA for the e-skin applications 34 .…”