Part of the show Cutting Edge in Cancer
International computer giant IBM has unveiled its vision for a "supercomputer in a sugarcube" powered by "electronic blood".
The aim is to base the design on the workings of the human brain in order to surmount some of the difficulties - delivering power and removing the excess heat - that computer engineers are currently encountering as they seek to make chips smaller and faster.
The heat problem effectively doubles the running cost of a computer, because nearly as much energy has to be spent on cooling as on computing.
Current figures suggest that just running the Internet accounts annually - in energy terms - for an equivalent amount of CO2 as the airline industry, and computing generally equates to a global spend of over $30 billion just making hot air.
But, to make processors more compact and run even faster, chips will need to shrink, partly to reduce the distances - and hence time - over which information needs to be transmitted and to pack in more processing power. Doing so, though, will intensify the present problems.
So IBM have turned to the human brain for inspiration because, despite consuming energy at the rate of only about 20 watts, the brain is, according to IBM researchers Patrick Ruch and Bruno Michel, "10,000 more dense and efficient than any computer around today."
This they put down to the fact that the brain solves several problems in one: the 3D structure of the brain makes it highly compact, and it uses a highly efficient circulatory (blood) system to deliver energy and keep the operating environment right for optimal performance of the components.
According to Michel, the brain uses "40% of its volume for function and 10% for energy and cooling." The best current computers, on the other hand, devote only about 1% to processing!
Instead, Michel and Ruch propose to pile up processors like the chip equivalent of a skyscraper to boost computing power; then, they'll keep them cool by circulating a fluid - "electronic blood" - through special channels within the chips to soak up heat; and - here's the really clever bit - they'll also use the same fluid to power the chips using a chemical reaction.
Analogous to blood delivering sugar to hungry neurones, the approach being explored at IBM is to use a solution of a chemical like vanadium that can be "charged" before it is pumped into the computer.
Then, as it passes over the processor components, the vanadium alters its chemical oxidation state, releasing electrons to the transistors to power them.
The ultimate goal is to cram a super-computer that currently occupies half a football pitch into a volume the size of a sugarcube by 2060.
Nobody really knows the future but I think that computers will be required to operate at least as well as human brains work. At present we don't know how the intelligence structure works or the mechanisms involved to allow us to solve certain kinds of problems and so our intelligence cannot yet be pinned down to a mechanistic process. However, it is known that inside the neurons there are microtubules which contain very special type of pure water which may act so as to insulate the wavefunctions of quantum electrons, thus preserving their wavefunctions. The idea is that before every neuron signal occurs there are vast numbers of quantum calculations going on, whose final output either fires the neuron or doesn't (binary fashion). But exactly how this quantum computing works is very much speculation. With quantum entanglement there are obviously very many more paths in the network than is suggested by the number of neuron interconnections alone. The best people to answer this question is probably the quantum-computer designers. Does any such person read this forum? Would they like to enlighten us? However, bear in mind that even IBM doesn't know how intelligence is done, because if they did, we would have intelligent computers by now, wouldn't we. I know that some people will point out that the wavefunction above will be lost as the wavefunction comes into contact with the water molecules. Based on this objection I don't know how this quantum computer is supposed to work, but it was suggested in the first place by Professor Sir Roger Penrose after Frohlich's 1975 idea for the possibility of quantum coherence phenomena in biological systems. It requires that "for quantum coherence to occur, the energy of the biological drive must be high enough before any coherence could be comparable in size to the large-scale behaviour shown in superconductivity". Apparently "there already exists evidence of oscillations occuring at 10 exp 11Hz in such biological structures". Ref Shadows of the Mind, page 367, chapter on quantum coherence in biological systems. The book was first published in 1995 so it may well be out of date by now and further advances may well exist. woolyhead, Sat, 2nd Nov 2013