Today’s computing are not that efficient for parallel pattern recognition and are struggling up with respect to energy efficiency and bandwidth requirement. Von Neumann architecture consume lot of power and require an address for further processing. The traditional approach is to continue pushing the limits of today’s systems and chips. However, another way is to incorporate the old technique that was conceived in 1980s- neuromorphic computing that is heating in recent years.
Neuromorphic computing utilizes computational functional of brain for processing data in the memory. Neuromorphic chips are fully parallel silicon neural network that has identical neurons for storing and processing information. The neuromorphic chips are basically pattern matching engines that enable systems to perform various tasks including machine learning, computer vision among others.
Companies involved in this arena:
- General Vision, fabless IC design house and supplier of neuromorphic chips
- General Vision has licensed its technology to Intel having 128 neurons and 128 bytes of memory per neuron
- IBM has also entered into the market launching TrueNorth
- Various universities and governmental agencies such as European Consortium, U.S. Department of Defense are working on it
Neuromorphic Chip Market: An architecture that mimics the human brain functionality
- Definition & Scope
- Key research’s in Neuromorphic Engineering (MIT, Spintronic Researchers, HP Labs, Stanford University and others)
- Upsurge in the adoption of very-large-scale integration (VLSI) systems
- Optimize utilization of power compared with normal PCs
- Research initiatives undertaken by major manufacturers
- Key Applications (Image recognition, Signal recognition, Data mining)
- Impact of Neuromorphic Chip in various industries
- Technological Changes in the Road Ahead