IBM’s Revolutionary TrueNorth Chip Inches Closer to Mimicking Human Brain

The launch of IBM’s new TrueNorth chip, containing 5.4 billion transistors, signals an evolution toward computing modelled after the functionality of the human brain.

However, there is much more R&D required, and according to analysts, even with multicore processors and upgraded memory storage, new computing technology falls short when compared to the computational power of the human brain. For example, the human brain has trillions of synapses. By comparison, even a neural network requiring 96 Lawrence Livermore National Lab Sequoia supercomputers delivers but a fraction of a human brain’s processing power, at a speed of 1,500 times less than the human brain. But, while a computer cannot match the computing power of a human brain at present, they are able to perform similar calculations and tease out complex reactions. To accomplish these functions though, computers consume vast amounts of energy.

Now, TrueNorth chips position IBM closer to realizing the possibility of computing modelled on the complexity of the human brain. 

Starting in 2008, one of IBM’s objectives was to produce a novel form of computing architecture patented on the nerve cells and neural networks of the human brain. The TrueNorth chip is a direct result of those efforts. Instead of simply performing calculations on multiple cores as quickly as possible, The TrueNorth chip handles data more effectively than other chips because of the chip’s 4,096 inner cores. Moreover, these cores are uniquely interoperable. Instead of all cores running at same time, each core activates only as necessary, resulting in more efficient power management and heat disbursement. Further, all cores are intelligently networked, resulting in scalability and minimal disruption/downtime.

Presently, these cores output a similar power quotient to a million nerve cells and 256 million synapses in the human brain. Ultimately though, the goal is for this chip to more closely mirror the brain more, capable of computing similarly large amounts of data on par with the human brain.

Source: Techtimes