Learn about INI's neuromorphic hardware: ROLLS

ROLLS At A Glance

Release Year: 2015
Status: End Of Life
Chip Type: Mixed-signal
Software: Custom Python
Applications: Research
Neurons: 256
Synapses: 64000
On-Chip Learning: true
Power: ~5 mW

ROLLS is a reconfigurable neuromorphic chip with 256 silicon neurons and 128K plastic synapses that implements spike-timing learning rules. It allows emulation of neural systems with adaptive behaviors using analog neuron/synapse circuits with added digital configuration logic for flexibility.

Developed By:


The ROLLS chip implements key requirements for online learning and adaptive behavior in neuromorphic systems, with highly flexible digital configuration options.


  • Presented as a “Reconfigurable On-line Learning Spiking Neuromorphic Processor”
  • 256 silicon neurons, 128K plastic synapses, 256K programmable synapses
  • Mixed-signal analog/digital implementation
  • 180nm CMOS process, 51.4 mm2 die area
  • Modeling of neural systems that adapt in real-time through on-chip learning

Silicon Neurons

  • Implement adaptive exponential Integrate-and-Fire (I&F) model
  • Exhibit various spiking behaviors (frequency adaptation, refractory period, threshold adaptation, bursting)
  • Highly configurable through 13 tunable bias parameters
  • Low mismatch between neurons (9.4% avg)
  • On-chip learning circuits to evaluate spike-timing-based plasticity rules


  • Two core arrays:
    • Long-term plasticity (LTP) synapses with bi-stable weights using drift-diffusion mechanism
    • Short-term plasticity (STP) synapses with 4-level programmable weights and short-term depression
  • LTP array models NMDA receptors with longer time constants
  • STP array models AMPA/GABA receptors with shorter time constants
  • Sparse digital logic for memory and configuration in every synapse
  • Digital configuration of synapse types and connectivity


  • Virtual “diffuser” synapses to combine signals from groups of synapses
  • Synapse demultiplexing to allow flexible neuron-synapse mapping
  • Asynchronous peripheral digital circuits for input/output
  • Bias generator for tuning analog parameters
  • ADC readout circuit for monitoring internal signals


  • Hardware attractor networks to model working memory and decision making
  • Real-time image classification demonstration with spiking vision sensor
April 2015A reconfigurable on-line learning spiking neuromorphic processor comprising 256 neurons and 128K synapsesNing Qiao, Hesham Mostafa, Federico Corradi, Marc Osswald, Fabio Stefanini, Dora Sumislawska, Giacomo IndiveriFrontiers of Neuroscience
Spiking Neural Network (SNN) Library Benchmarks

Spiking Neural Network (SNN) Library Benchmarks

  • Gregor Lenz, Kade Heckel, Sumit Bam Shrestha, Cameron Barker, Jens Egholm Pedersen

Discover the fastest Spiking Neural Network (SNN) frameworks for deep learning-based optimization. Performance, flexibility, and more analyzed in-depth

TrueNorth: A Deep Dive into IBM's Neuromorphic Chip Design

TrueNorth: A Deep Dive into IBM's Neuromorphic Chip Design

  • Fabrizio Ottati

Explore the innovative TrueNorth neuromorphic chip, its event-driven architecture, low power operation, massive parallelism, real-time capabilities, and scalable design.

Efficient Compression for Event-Based Data in Neuromorphic Applications

Efficient Compression for Event-Based Data in Neuromorphic Applications

  • Gregor Lenz, Fabrizio Ottati, Alexandre Marcireau

Discover methods to efficiently encode and store event-based data from high-resolution event cameras, striking a balance between file size and fast retrieval for spiking neural network training.