IBM has tied its hands with the United States’ AFRL or the Air Force Research Laboratory for developing a new brain-inspired supercomputer. This computing device will be powered up by a line of 63 IBM TrueNorth chips, structures of which mimic the synaptic connections of the brains’ neurons. The AI inspired supercomputer will function like the human brain. Sources also say that the supercomputer will have the ability to perform on the scalable level and will also have an end-to-end ecosystem of the software that in turn will enable the deep neural network learning as well as the information of the discovery. The sensory processing power is equivalent to the 64 million neurons, along with 16 million synapses, which allow it to be much more efficient than the existing systems of the conventional chips.
More about the System of IBM AFRL
It is being said that the entire system will fit into a 4U space of rack of a standard server pack. Eight of the systems will allow the unprecedented scale of 512 million neurons in each of the racks. A single processor will be made up of nearly 5.4 million transistors which will be organized into 40956 neuronal nuclei that eventually create an array of one million different digital neurons to communicate with each other through 256 million electrical synapses. The brand new TrueNorth neuron synaptic device will be able to interpret the data from many sensors like videos, images, symbols, all that in real time efficiency. The AFRL will make it a point to combine the perceptual abilities that are similar to the functions of the human brain and also its data processing capabilities. This system will not only enable to scale the work of the system but will also create a parallelism for the flow of the data, which, in the paradigm of the programming world can uplift a great working condition for the network. It is now, a matter of time, to see when the supercomputer will be developed and how it creates a niche in the technological platform and also the defense line of America in the future.