The world of silicon based processors will end soon. Silicon served the computing world since the start, but it reaches its limits, at least in the present form. IBM invests $3 billion in research to come up with a worthy replacement. The present processing power growth rate largely respected Moore’s Law. Moore, Intel’s co-founder, predicted a two year period for the processing power to double. A while ago we wrote about Titan Super Computer who was the fastest one at the time. A post silicon computing era requires a fundamental change.
“The basic architecture of the computer has remained unchanged since the 1940s. We feel, given the kinds of problems we see today, [that] this is the time to start looking for new forms of computing,” declared Supratik Guha, director of physical sciences for IBM Research.
Intel will soon produce 14 nm processors. Ten nm versions are already planned. But going smaller than 7 nm will be quite a challenge, according to IBM. The company has several research directions to follow.
Post silicon computing might rely on carbon
The importance of having new prospects for chip development cannot be stressed enough. Scientists already dream on researching some of the most complex phenomena like weather patterns or cancer development. Huge amounts of processing power will be required by such projects. One of the main problem, though, is how to produce a more energy efficient chip. Today’s super computers use thousands of processing cores simultaneously and powering them up is very costly. Cooling them brings the same problem as well.
IBM will invest $3 billion dollars in the next 5 years to come up with a solution. In the meantime, strangely, it decided to sell its plants, which have not been particular financial successes. The division producing lower-end servers will soon be a part of Lenovo for $2.3 billion.
Carbon nanotubes are potential replacers of silicon once the cooling problem is solved. But silicon might not be totally abandoned. The silicon nanophotonics concept relies on light to transmit data instead of electricity. Quantum computing will be a viable option once the hardware will be able to support it. Qubits lie at the core of quantum computing. The difference between bits and qubits is the fact that qubits can be assigned values of 1 and 0 at different states, compared to bits who are only working in one state. When researchers manage to find a way to train the qubits to maintain desirable states, the era will begin. Now that IBM Invests $3 Billion in the process, we might use post silicon computing in the near future.