How to replicate the squishy sophistication of the human brain in hard metal and silicon? IBM thinks it鈥檚 found a way, and to prove it has two new 鈥渃ognitive computing鈥� microchips whose design is inspired by the human brain.
In the mammalian brain, neurons send chemical signals to each other across tiny gaps called synapses. A neuron鈥檚 long 鈥渢ail鈥�, the axon, sends the signals from its multiple terminals; the receptive parts of other neurons 鈥� the dendrites 鈥� collect them.
Each of IBM鈥檚 brain-mimicking silicon chips is a few square millimetres in size and holds a grid of 256 parallel wires that represent dendrites of computational 鈥渘eurons鈥� crossed at right angles by other wires standing in for axons. The 鈥渟ynapses鈥� are 45-nanometre transistors connecting the criss-crossing wires and act as the chips鈥� memory; one chip has 262,144 of them and the other 65,536. Each electrical signal crossing a synapse consumes just 45聽picajoules 鈥� a thousandth of what typical computer chips use.
Advertisement
Because the neurons and synapses are so close together, the pieces of hardware responsible for computation and memory are also much closer than in ordinary computer chips. Conventionally, the memory sits to the side of the processor, but in the new chips the memory 鈥� the synapses 鈥� and the processors 鈥� the neurons 鈥� are on top of each other, so they don鈥檛 need to use as much energy sending electrons back and forth. That means the chips can perform parallel processing far more efficiently than conventional computers.
In preliminary tests, the chips were able to play a game of Pong, control a virtual car on a racecourse and identify an image or digit drawn on a screen. These are all tasks computers have accomplished before, but the new chips managed to complete them without needing a specialised program for each task. The chips can also 鈥渓earn鈥� how to complete each task if trained.
Fewer watts than Watson
Eventually, by connecting many such chips, of IBM Research 鈥� Almaden, in San Jose, California, hopes to build a shoebox-sized supercomputer with 10聽billion neurons and 100聽trillion synapses that consumes just 1聽kilowatt of power. That may still sound a lot 鈥� a standard PC uses only a few hundred watts 鈥� but a supercomputer like IBM鈥檚 Watson uses hundreds of kilowatts. By contrast, the ultra-efficient human brain is estimated to have 100聽billion neurons and at least 100聽trillion synapses but consumes no more than 20聽watts.
of Stanford University, California, says scale is one of the key issues. Until the chips contain as many synapses as the human brain, it will be difficult to distinguish their accomplishments from those of other computers.
The chips are sponsored by a US Defense Advanced Research Projects Agency (DARPA) project to create computers whose abilities rival those of the human brain.
