POETS: A Parallel Cluster Architecture for Spiking Neural Network

Spiking Neural Networks (SNNs) are known as a branch of neuromorphic computing and currently used in neuroscience applications to understand model the biological brain. SNNs could also potentially be many other application domains such classification, pattern recognition, autonomous control. This work presents highly-scalable hardware platform called POETS, uses it implement SNN on very large number parallel reconfigurable FPGA-based processors. The current system consists 48 FPGAs, providing 3072 processing cores 49152 threads. We use this up four million neurons with one thousand synapses. Comparison similar platforms shows that POETS is twenty times faster than Brian simulator, at least two SpiNNaker.