By rapidly reconfiguring to execute different portions of each function, a 3D Spacetime device can implement a complex design using only a small fraction of the resources that would be required by an FPGA, with its inherently 2D architecture.” The big plus is the manageable price-point of Spacetime devices.
Beyond programmability, self-learning processors. Programmable hardware is the future but cognitive computing is even more exciting. You will be surprised to know that, sometime this year, computers that learn from their own mistakes will be commercially available!
Stanford University, IBM, Google and several others are working on this. Such self-learning computers will enable extreme automation, fault tolerance and adaptability, making computer crashes a thing of the past. They will use a new class of processors, which comprise electronic components that can be connected by wires that mimic biological synapses. Because of the neuron-like elements in these processors, they are known as ‘neuromorphic processors.’ These processors need not be programmed; instead the connections in the circuits are formed based on correlations in data that the processor has learned earlier. As data flows, the connections might get further altered, leading to changes in behaviour just as in the case of humans.
According to a comment by IBM scientist Dharmendra Modha, this methodology brings computation to data instead of the other way round. “Sensors become the computer, and it opens up a new way to use computer chips that can be everywhere,” he said in an earlier report. When they first arrive, these computers will work as co-processors in devices such as smartphones rather than as standalone processors. Qualcomm, which has developed such a processor together with IBM, plans to release it commercially sometime this year. Stanford’s Brains in Silicon project is another major contender in this space.
The author is a technically-qualified freelance writer, editor and hands-on mom based in Chennai
