Meebo Chat Bar

Sunday, September 6, 2009

Future directions in computing

Future crunch: Six advances that could spell the end of traditional computing

* Quantum
* Light
* Spin
* Chemicals
* DNA
* Plastics

Quantum

Silicon electronics are a staple of the computing industry, but researchers are now exploring other techniques to deliver powerful computers.
Quantum computers are able to tackle complex problems

Quantum computing graphic
A quantum computer is a theoretical device that would make use of the properties of quantum mechanics, the realm of physics that deals with energy and matter at atomic scales.
In a quantum computer data is not processed by electrons passing through transistors, as is the case in today's computers, but by caged atoms known as quantum bits or Qubits.

"It is a new paradigm for computation," said Professor Artur Ekert of the University of Oxford. "It's doing computation differently."

A bit is a simple unit of information that is represented by a "1" or a "0" in a conventional electronic computer.
A qubit can also represent a "1" or a "0" but crucially can be both at the same time - known as a superposition.

This allows a quantum computer to work through many problems and arrive at their solutions simultaneously.

"It is like massively parallel processing but in one piece of hardware," said Professor Ekert.

'Complex systems'

This has significant advantages, particularly for solving problems with a large amount of data or variables.

"With quantum computing you are able to attack some problems on the time scales of seconds, which might take an almost infinite amount of time with classical computers," Professor David Awschalom of the University of California, Santa Barbara told the BBC News website recently.

In February 2007, the Canadian company D-Wave systems claimed to have demonstrated a working quantum computer.

At the time, Herb Martin, chief executive officer of the company said that the display represented a "substantial step forward in solving commercial and scientific problems which, until now, were considered intractable."

But many in the quantum computing world have remained sceptical, primarily because the company released very little information about the machine.

The display also failed to impress.

"It was not quite what we understand as quantum computing," said Professor Ekert. "The demonstrations they showed could have been solved by conventional computers."

However, Professor Ekert believes that quantum computing will eventually come of age.

Then, he said, they will not be used in run-of-the-mill desktop applications but specialist uses such as searching vast databases, creating uncrackable ciphers or simulating the atomic structures of substances.

"The really killer application will probably be in designing new materials or complex systems," he said.


Light


Computers exploit the movement or accumulation of electrons to do useful calculations. These movements down tiny wires are the reason why PCs become so hot.

Fibre optics
Optics are already used to transfer data over the internet

"We are dissipating huge amounts of power in chips right now," said Professor Stan Williams of computer firm Hewlett Packard.

And the problem will get worse as the components become smaller, making chips even more inefficient, he said.

"If we could find new ways of moving information around a chip, we could significantly reduce the amount of power that is dissipated."

A potential solution would be to use particles of light - photons - instead of electrons to move information around and between chips.

"Electrons will still be doing the computing but photons will be doing the communicating," he said.

"It sounds a little bit complex but you find that that baton pass [of data between the electrons and photons] can be far more efficient, in terms of electrical power required than just using electrons."

It is similar to the technology that is used to move data around the internet today but at a much smaller scale.

The technology could be key for transferring data between massively multiple core chips - devices with several linked computing engines.

But some researchers would ultimately like to skip this baton pass and use photons to manipulate and store data, rather than just transmit it.

Experiments are going on in academic institutions, firms such as IBM and even the American space agency Nasa. Several of the individual components needed to build an optical computer have been demonstrated and even put together into a working machine.

However, it may be some time before optical components compete with silicon because of a fundamental barrier.

"The reason it is not in circuits today is essentially cost," said Professor Williams.


Spin


Spintronics, also known as magnetoelectronics, is a technology that harnesses the spin of particles, a property ignored by conventional electronics.

Concept picture of spinning electron
Spintronics harnesses the spin of sub-atomic particles

"Until now, electronics has worked by moving electrons around or moving charge around and that takes work," said Kevin Roche of computer giant IBM. "The most obvious example of that is that if you have a laptop that runs faster, it runs hotter."

But, by using the spin of particles - detected as a weak magnetic force - scientists believe they can unlock almost infinite computing power.

"It is called spin because the maths for dealing with it is similar to the maths for a spinning ball," said Mr Roche. "An electron always has spin and it can be spinning one of two different ways: up or down."

These two different states can be used to represent a "1" or a "0" - the bits of information used by all computers.

Basic spintronic devices are already used in today's computers.

For example, most hard drives today use a "spin valve", a device that reads information off the individual disks or platters that make up a hard drive.

But researchers at firms such as Intel and IBM are hoping to take this one stage further.

"We are hoping to use it to store data, to transmit data; to do all the things we do now with charge but to use the spin property of the electron instead," said Mr Roche.

Chips exploiting spintronics would in theory be able to transmit data with spin using a lot less charge.

"So power costs go down," said Mr Roche.

IBM has already shown off a prototype spintronic device known as "racetrack memory", a device that could increase storage density by up to 100 times.

Other researchers are working on spin-based transistors, which unlike conventional transistors would not require the application of an electric current to work.

According to many researchers, spintronics could be the next big change in computing in the coming decades.

"The conventional microelectronics industry and the magnetic storage industry are approaching their limits very fast. Spintronics might offer a way out," Dr Yongbing Xu of the University of York told the BBC earlier this year.

Chemical Computing

Chemical computing is an unconventional approach to computation that uses a "soup" where data is represented by different concentrations of chemicals

Diffusion in a beaker
Chemcial computing can be used to solve a range of problems

Chemical computers can exploit several different kinds of reaction to carry out the computation.

For example, so-called conformation computers use polymer molecules that change shape in response to a particular input. Metabolic computing exploits the kinds of reactions typically found inside a living cell.

Dr Andrew Adamatzky of the University of West England works on another type.

"I am dealing only with reaction-diffusion computing," he explains.

This type of computation exploits waves travelling through a beaker of chemicals to carry out useful calculations.

These waves are the information carriers in the computer. They are created by triggering chemical reactions in the soup at specific points.

As waves propagate from different areas they collide and interact - effectively processing the information they hold. At the site of their interaction a point with a new chemical concentration is created, which is in effect an answer.

With a beaker full of thousands of waves travelling and interacting with each other, complex computational problems can be solved.

Robot gel

Although the process sounds complicated and esoteric it can be applied to almost all computational problems.

"Reaction-diffusion processors are universal computers, they can solve all types of problems," said Dr Adamatzky.

As a result, computer giant IBM is already interested in the technology.

Although slower than silicon, its key advantage is that it is cheap to produce and incredibly robust.

Working with chemist Ben De Lacy Costello, Dr Adamatzky has already produced logic gates using the technique that can be used to make chemical "circuitry".

"Ultimately, we will produce a general purpose chemical chip," he said.

The chip would be capable of mathematical operations such as adding and multiplying numbers, he said.

However, he believes he can take the research even further to create intelligent, amorphous robots.

In these, he said, silicon circuitry would be of no use

"Assume we have fabricated an artificial amoeba, gel-based robot, without any fixed shape, and capable for splitting into several smaller robots," he said.

"Conventional silicon circuits will not work because they have rigid architecture."

But as chemical computers are an amorphous blob they could be cut in half and both would continue functioning independently.

"You can not cut your laptop in half and expect both parts to function properly; you can do this with reaction-diffusion processors," he said.


DNA computing


DNA computing, commonly called biomolecular computing, is an emerging field that uses DNA and biochemistry instead of silicon-based electronics.

DNA
Professor Shapiro has designed a DNA computer to target cancer

The first proof-of-concept use of DNA to perform computation was carried out by Professor Leonard Adleman at the University of Southern California in 1994.

The original goal of the field was to use biomolecules to beat electronic computers at solving large complex problems.

"Today, most people believe that biomolecular computing will not beat electronic computers in the foreseeable future," admitted Professor Ehud Shapiro of the Weizmann Institute of Science in Israel.

Instead, said Professor Shapiro, the new goal is to try to use DNA computing to do things that traditional silicon cannot.

"What we offered in our work was a different vision," he said.

In particular he is trying to develop smart drugs that are capable of computation.

"You want a drug that can sense the biochemical environment, analyse it and, in response, release a molecule which is the appropriate drug for that particular situation," he said.

Cancer buster

In 2002, Professor Shapiro unveiled a programmable molecular device composed of enzymes and synthetic strands of DNA.

Two years later, the same team showed off another DNA computer so small that roughly a trillion of them could fit into a microlitre (a millionth of a litre).

The device was able to detect signs of cancer, and release drugs to treat the disease.

"This soup of DNA and enzymes implements a well know mathematical model of computation known as finite automaton," he explained.

"This finite automaton knows how to do very simple computation such as recognising whether a list of zeros and ones has an even number of ones."

In the case of his 2004 computer this method of computation was used to analyze ratios of specific molecules related to prostate cancer and a specific type of lung cancer.

The "computer" consisted of a chain of three segments of DNA and an enzyme which could cut the strands.

If the first segment of the DNA detects ratios of certain molecules that indicate the presence of cancer, it tells the second segment to release the third segment, which is an anti-cancer drug.

Since then, other researchers have shown off new computational systems that make use of enzymes that naturally occur in a living cell, whilst Professor Shapiro has attempted to make a more practical device.
"We have continued working in this direction trying to make this molecular system to work in living cells," he said.

Plastic

Silicon is expensive and complex to produce, requiring clean rooms and precise manufacturing techniques in plants that can cost billions of dollars.

Flexible television
Organic electronics can be used to make flexible displays

This means they are currently out of reach for low-cost products.

But organic polymers, a class of substances that are used to make everything from bin bags to solar panels, could offer a solution.

"It really opens up a whole new set of options for what you do with electronics," said Professor Art Ramirez of Bell Labs.

Highly conductive polymers were first discovered in the early 1960s and are already used in some electronic devices.

In 2004, electronics giant Philips announced a concept flexible display, while other companies such as Cambridge Display Technology use them to manufacture organic light-emitting diodes (LEDs).

Earlier this year, UK firm Plastic Logic said that it would build the world's first factory to produce plastic electronic circuits.

Although, circuits made of polymers are much slower than silicon devices, they have one significant advantage: they can be printed using techniques similar to those used to mass produce magazines and wallpaper.

"Using big drums of substrate you can put down one chemical after the other and build up something that looks like a circuit," said Professor Ramirez.

This means they are cheap and easy to produce.

And as the polymers can be printed onto flexible substrates they can also be used in totally new types of device, such as large-scale billboards or electronic newspapers.

"If you can integrate that with very simple wireless connectivity one can imagine, instead of picking up the New York Times each morning, picking up a plastic sheet," said Professor Ramire

http://news.bbc.co.uk/2/hi/technology/7085019.stm