A large-scale spiking neural networks emulation architecture

The purpose of this work is to design a new version (called SNAVA+) of the architecture SNAVA, an SNN hardware emulator implemented on a XilinX Kintex-7 FPGA. SNAVA+ increases the capabilities of SNAVA in order to have a large-scale SNN emulator. It preserves the basic hardware structure of SNAVA, m...

Descripción completa

Detalles Bibliográficos
Autor: Pirrone, Vito
Tipo de recurso: tesis de maestría
Fecha de publicación:2014
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2099.1/22996
Acceso en línea:https://hdl.handle.net/2099.1/22996
Access Level:acceso abierto
Palabra clave:Neural networks (Computer science)
Spiking Neural Networks
SNN hardware emulator
hardware architecture
FPGA
Xarxes neuronals (Informàtica)
Àrees temàtiques de la UPC::Enginyeria de la telecomunicació
Descripción
Sumario:The purpose of this work is to design a new version (called SNAVA+) of the architecture SNAVA, an SNN hardware emulator implemented on a XilinX Kintex-7 FPGA. SNAVA+ increases the capabilities of SNAVA in order to have a large-scale SNN emulator. It preserves the basic hardware structure of SNAVA, making however several changes to optimize the performance. In particular, SNAVA+ project focuses on the aim to exploit more efficiently the availables resources, to reduce both the area and power consumption of the FPGA. A better use of the resources, in fact, is the main key to increase the potentiality of the SNN emulator, i.e. to increase the number of neurons and synapses simulated.