New week of posts. I am coming back today to the definition series from Techtarget glossary website WhatIs.
Neurosynaptic chips are the hardware counterpart of the software based cognitive computing. Emulating the brain processes as they are on the biological level have proved to be very hard for artificial mechanisms. Neurosynaptic chips promise to emulate the functioning of the human brain cells with better efficiency than current approaches. The key to accomplish the required efficiency is a precise combination of distributed and parallel computing processors. IBM is the company most involved in this effort, comprehensible trying to bind neurosynaptic chips with their major accomplishment – the Watson cognitive computing machine software best of breed.
A neurosynaptic chip, also known as a cognitive chip, is a computer processor that functions more like a biological brain than a typical CPU does. Unlike cognitive computing, which is made to emulate the thought and learning of humans through software, neurosynaptic chips are made to function like human brains on the hardware level.
While a typical computer works well for language, mathematical and data analytics processing, it can take a lot of work for it to perform tasks that even simple biological brains are efficient at. A neurosynaptic chip is more efficient at these tasks, which include pattern recognition and sensory processing and learning.
Here’s how IBM describes the new architecture:
“IBM’s brain-inspired architecture consists of a network of neurosynaptic cores. Cores are distributed and operate in parallel. Cores operate—without a clock—in an event-driven fashion. Cores integrate memory, computation, and communication. Individual cores can fail and yet, like the brain, the architecture can still function. Cores on the same chip communicate with one another via an on-chip event-driven network. Chips communicate via an inter-chip interface leading to seamless availability like the cortex, enabling creation of scalable neuromorphic systems.”
Indeed the functioning of the Human Brain is more akin to an unstructured jam of signaling between organic matter cells (nodes), instead of a precise clocked structured processor core, like our computer processing core functions. The application space for this new technological paradigm may be a huge one, promising to render the computational and energy costs of current supercomputer implementations an obsolete paradigm. At the same time, and following the insight of the Human Brain as a super energy-efficient supercomputer, this new technology would be a decisive contribution to the efforts in a fully complete understanding of the Human Brain:
Especially in mobile technology, where processing and power are limited, the neurosynaptic chip stands to revolutionize abilities. Tasks like selecting the best produce or finding repair points in electronics could be carried out using smartphone cameras. For traditional computer architectures, it takes a supercomputer to perform these tasks, consuming massive amounts of power in the process. Neurosynaptic chips make this possible with a tenth of the energy requirements.
The new type of chip also has promise for supercomputing applications. IBM has a neurosynaptic processor project, called Brainpower. MIT has simulated a functioning brain synapse in their quest for truly intelligent systems. A goal of the IBM project is a trillion synapses with only 4kW.
Currently in its 2nd generation, the neurosynaptic chip has made impressive strides in its specs:
- Generation 1 – 256 programmable neurons, 262114 programmable synapses, 1 neurosynaptic core
- Generation 2 – 1 million programmable neurons, 256 million programmable synapses, 4096 neurosynaptic cores