Toward achieving one million times increase in computing efficiency

Jul 10, 2012

Modern-day computers are based on logic circuits using semiconductor transistors. To increase computing power, smaller transistors are required. Moore's Law states that the number of transistors that can fit on an integrated circuit should double every two years due to scaling. But as transistors reach atomic dimensions, achieving this feat is becoming increasingly difficult. Among the most significant challenges is heat dissipation from circuits created using today's standard semiconductor technology, complementary metal-oxide semiconductor (CMOS), which give off more heat as more transistors are added. This makes CMOS incapable of supporting the computers of the future.

Engineers are therefore seeking alternatives to CMOS that would allow for highly efficient circuits that generate much less heat. Northwestern University researchers may have found a solution: an entirely new logic circuit family based on magnetic . The advance could lead to logic circuits up to 1 million times more power-efficient than today's.

Unlike traditional integrated circuits, which consist of a collection of miniature transistors operating on a single piece of semiconductor, the so-called "spin logic circuits" utilize the phenomenon of spin, a fundamental property of the electron.

"What we've developed is a device that can be configured in a logic circuit that is capable of performing all the necessary Boolean logic and can be cascaded to develop sophisticated function units," said Bruce W. Wessels, Walter P. Murphy Professor of Materials Science and Engineering at Northwestern's McCormick School of Engineering and Applied Science and one of the paper's authors. "We are using 'spintronic' logic devices to successfully perform the same operations as a conventional CMOS circuits but with fewer devices and more computing power."

The spin- are created with magnetoresistive bipolar spin-transistors, recently patented by the researchers.

A paper describing the findings, "Emitter-Coupled Spin-Transistor Logic," was presented July 5 at the International Symposium on Nanoscale Architectures held in the Netherlands. Additional Northwestern authors include graduate student Joseph Friedman, the paper's lead author; Gokhan Memik, associate professor of electrical engineering and computer science; and Alan Sahakian, professor of electrical engineering and computer science.

The new logic family, which takes advantage of the magnetic properties associated with electron spin, could result in a computer 1 million times more power-efficient than those on the market today. While that achievement is optimistic and could take a decade to realize, "We think this is potentially groundbreaking," Friedman said.

Explore further: Atoms to product: Aiming to make nanoscale benefits life-sized

Related Stories

Memristor chip could lead to faster, cheaper computers

Mar 17, 2009

(PhysOrg.com) -- The memristor is a computer component that offers both memory and logic functions in one simple package. It has the potential to transform the semiconductor industry, enabling smaller, faster, cheaper chips ...

Physicists put a new spin on electrons

Apr 15, 2009

In the first demonstration of its kind, researchers at the University of British Columbia have controlled the spin of electrons using a ballistic technique--bouncing electrons through a microscopic channel of precisely constructed, ...

Recommended for you

Biomimetic photodetector 'sees' in color

Aug 25, 2014

(Phys.org) —Rice University researchers have created a CMOS-compatible, biomimetic color photodetector that directly responds to red, green and blue light in much the same way the human eye does.

User comments : 25

Adjust slider to filter visible comments by rank

Display comments: newest first

210
1.7 / 5 (3) Jul 10, 2012
"The advance could lead to logic circuits up to 1 million times more power-efficient than today's." Indeed.... Size, did I miss it or did the article mention supporting the gate count increase with a concomitant size reduction in gates and interconnect?
To support Moore's Law and all...

word-to-ya-muthas
patnclaire
not rated yet Jul 10, 2012
Sounds good until you have to pass the PC through TSA airport security screens including X-rays, T-rays, and magnetic wands.
Bowler_4007
1.3 / 5 (3) Jul 10, 2012
recently patented by the researchers
could take a decade to realize
anyone wondering why its gonna take so long? why did they have to patent it? if i'm not mistaken technological advances such these spin logic circuits are supposed to advance our societies whereas these people have pulled out a gun and said "if you want to make use of this technology then pay us".

gotta say say i'm happy about the advance but not with these people
Deesky
4 / 5 (7) Jul 10, 2012
anyone wondering why its gonna take so long?
Not me. Do you have any idea how expensive chip manufacturing facilities are? The investments made in these plants are staggering! You cannot just one day decide to do something totally different - it takes time, research, testing and investment before you can even think about setting up a new commercial plant.

why did they have to patent it?
if i'm not mistaken technological advances such these spin logic circuits are supposed to advance our societies whereas these people have pulled out a gun and said "if you want to make use of this technology then pay us".

The world simply does not work like that.

gotta say say i'm happy about the advance but not with these people

There is absolutely nothing wrong with 'these' people. That's how R&D works. If it wasn't for patents, you wouldn't be reading this on a computer.
Sure_about
not rated yet Jul 11, 2012
Don't spin waves only travel at around 10^5 meters per second? Won't that affect performance?
alfie_null
not rated yet Jul 11, 2012
Moore's Law states ...

Regarding density, still got increasing quantum tunneling issues, and - well - the atomic nature of atoms. Can't make 'em any smaller.

In Alfie's science writing style guide, use of the term is accompanied by a mandatory dope smack. It grates.

... one million times increase ...

Alfie's law of exponential hyperbole states that large quantities in predictions increase by an order of magnitude every 18 months.
DaFranker
3.1 / 5 (8) Jul 11, 2012

The world simply does not work like that.

(...)

There is absolutely nothing wrong with 'these' people. That's how R&D works. If it wasn't for patents, you wouldn't be reading this on a computer.

False on both counts. You are equating "the world" with "ideal american capitalism" and "R&D" with "currently dominant corporative policy regarding breakthroughs in technological development and design".

You're committing an implicit False Dilemma fallacy if you assume that the system you're defending is the only alternative to "anarchy" or whichever other scapegoat you posit implicitly. There *are* alternative options, some of which may be outright superior in every respect. Just because you refuse to consider the possibility doesn't mean absolute impossibility.

You're also turning the (centuries-old) response to a utility-maximizing function into Moral Truth.
DuncanI
2.5 / 5 (6) Jul 12, 2012
"There is absolutely nothing wrong with 'these' people. That's how R&D works. If it wasn't for patents, you wouldn't be reading this on a computer."

Untrue. Patents are legalised monopoly and have nothing to do with R&D and are basically a land grab. They are a hindrance to good products (features are missed off because of patent disputes.) Drug companies are an exception; they need a better system to fund development.
Bowler_4007
1 / 5 (1) Jul 12, 2012
Not me. Do you have any idea how expensive chip manufacturing facilities are?

i would imagine on the order of £100-million perhaps upto £1-billion, something like this would revolutionize the computing world though so it'd be worth it.

perhaps i jumped the gun when i ranted on about patents, do the companies/people that take out the patents get to decide how much is charged for use of the ideas within it?

i'm pretty sure i would be able to read this on a computer without patents, sure the tech would be a little different perhaps a bit backwards but computers aren't that complex and in fact without patents more people would have the right to construct the components in any way they liked the details of how to do so wouldn't have to exist in a patent i'm pretty people would have still wrote books to share information

btw these is a proper word so why use quotation marks on it?
Bowler_4007
2 / 5 (4) Jul 12, 2012
@Deesky, you know after reading some other comments i remember why i'm so against patents when it comes to fundamental technology, as was said earlier "legalised monopoly".

The world is obsessed with money and so are you, the first thing you jumped to was the cost of facilities to produce this stuff.

i don't much care for the way "the world" (your words not mine, it should be "society") works i reckon people could have still built technologically advanced societies even if we built our homes in a forest, people seem to forget that we're animals too and have as much right to the earth as every other creature but what do we do? we suck the life out of everything for money and power and honestly sometimes it makes me feel sick.

The only thing that stops me going insane is that i know that some people do care and want to see a totally different type of society to come about.

The only thing i hate as much as monopolised society is religion but i'm not gonna go into that.
Bowler_4007
1 / 5 (1) Jul 12, 2012
Don't spin waves only travel at around 10^5 meters per second? Won't that affect performance?
i think you mean spin current, i could be wrong because i haven't heard of spin waves before, anyway i think the propogation of a spin current might be determined by the inverse square law of force interaction of one electron to the next depending on their proximity, as they get closer their interaction gets stronger and the propogation should become faster and faster, i can't give an exact speed or even a range of speeds and i hope i got my facts right, but theres my answer based on my current understanding

anyone want to tell me if i got that right?
Deesky
4.3 / 5 (4) Jul 12, 2012
The world is obsessed with money and so are you, the first thing you jumped to was the cost of facilities to produce this stuff.

You have no idea what my obsessions are, but I will tell you. It's science, in all areas of applicability, including social science, psychology and self-organized social systems like cooperation and economics. So, yes, when examining a situation, you have to take all factors into account, and in this case, cost is a big factor.

i don't much care for the way "the world" (your words not mine, it should be "society") works

That much is obvious. You seem to want to live in your own little world, which is either extreme idealism or escapism. Either way, why get upset when people actually talk about how things really do work? If you wish to affect change, you need to understand reality and then do what you can to change it.
Deesky
3.7 / 5 (3) Jul 12, 2012
i reckon people could have still built technologically advanced societies even if we built our homes in a forest

I don't see what this has to do with anything under discussion. But, in any case, you are wrong. For a technological society to arise you need a high density of people living in close proximity where cooperation, trade, commerce and knowledge can be easily exchanged, initially by means of barter and later by inventing a more abstracted and sophisticated system - money.

Over time these societies grow, both culturally and technologically. Eventually the fruits of these societies make possible things like computers, telecommunications, the internet and rapid transportation of goods and people. All of which eventually make it POSSIBLE for those that wish, to live in forests!

OTOH, maybe you'd prefer to live in forests as some jungle native people still do? If so, you won't be doing it with any technologically advanced toys.
Deesky
5 / 5 (2) Jul 12, 2012
The only thing that stops me going insane is that i know that some people do care and want to see a totally different type of society to come about.

I assume you mean a more eco-friendly society? The only way that can happen is through greater education and continued technological and economic prosperity.

There is a strong correlation between advanced liberal democracies and conservation and sustainability advocacy. But, even that can only slow the rate of environmental exploitation if global population growth continues to increase at high levels.

Still, those advanced liberal democracies tend to have lower birth rates and some even below the replacement rate...
Bowler_4007
1.5 / 5 (2) Jul 13, 2012
i reckon people could have still built technologically advanced societies even if we built our homes in a forest

I don't see what this has to do with anything under discussion. .......

i don't think you understood and even if you did you didn't give a single reason as to why bulding a technologically advanced society in a forest would not be possible.

The only thing that stops me going insane is that i know that some people do care and want to see a totally different type of society to come about.

I assume you mean a more eco-friendly society?
nail on the head there.

i don't much care for the way "the world" (your words not mine, it should be "society") works

That much is obvious. You seem to want to live in your own little world, which is either extreme idealism or escapism.
i would like to live in a more eco-friendly society, i can however tolerate this society just enough to not go insane, so theres no extremism one way or another
NotParker
3 / 5 (2) Jul 14, 2012
Not me. Do you have any idea how expensive chip manufacturing facilities are?

i would imagine on the order of 100-million perhaps upto 1-billion,


5 billion US dollars plus.

http://www.tomsha...545.html
Eikka
1 / 5 (1) Jul 14, 2012

5 billion US dollars plus.


That's actually the reason why the Moore's law will stop before it becomes physically impossible.

Each new, better, fabrication technology is progressively more expensive to implement which means that each new chip has to sell more than the previous one to offset the cost. Otherwise the price per unit will go up and people can't afford to buy them anymore. At some point however, you'll run out of customers.

At the moment there are only a handful of different companies that can afford to make powerful processors. The number will be less with each technology generation, until it simply takes too much resources to make the next better one.

Because cost isn't just abstract money, but money represents the real drain in resources, materials and energy, that goes into making the products. Even without money, you'd have the exact same problem - only you'd have no way to measure how large the problem is.
SteveL
5 / 5 (1) Jul 14, 2012
And before millions of investors risk billions of dollars there had better be a patent on the technology or the investment simply won't be made.

It's not really just about greed, it's more about protecting an investment. If it's not protected by patents, someone will steal the technology after all the work has been done and the billions spent and those who made the investment will lose what they risked.
Lumberjack
not rated yet Jul 15, 2012
Deesky:
There is absolutely nothing wrong with 'these' people. That's how R&D works. If it wasn't for patents, you wouldn't be reading this on a computer.

@Deesky,
Do you have any evidence to support your assertion about the effect of patents (e.g., some scientific study comparing the effect on the technologic development just before and after the introduction of patents rights) or it's just some sort of pseudoreligious, unfounded belief?
88HUX88
not rated yet Jul 16, 2012
You said "Moore's Law states that the number of transistors that can fit on an integrated circuit should double every two years due to scaling".
No it doesn't, the paper merely described a trend. There has been surprise that the trend has lasted this long.
It remains to be seen if this potentially groundbreaking achievement materializes in commercial form.
antialias_physorg
not rated yet Jul 16, 2012
That's actually the reason why the Moore's law will stop before it becomes physically impossible. Each new, better, fabrication technology is progressively more expensive to implement which means that each new chip has to sell more than the previous one to offset the cost.

I'm not really convinced that is the case. The manufacturing cost per chip are peanuts. The development cost and setting up an assembly line are huge - but pale quickly into insignificance when compared to the profit made by even a modest amount of sales. Development costs would have to rise enormously before they'd impact the bottom line to any noticeable extent.

ROI is another matter. If the ROI isn't fast enough then investors will go somewhere else. Investors never have (and never will) care what product they invest in as long as the money comes in as fast as possible (that's why so many invest into weapons manufacturing)
SteveL
1 / 5 (1) Jul 16, 2012
ROI is another matter. If the ROI isn't fast enough then investors will go somewhere else. Investors never have (and never will) care what product they invest in as long as the money comes in as fast as possible (that's why so many invest into weapons manufacturing)
If this were the case there would be relatively few very large companies and it would be practically impossible for new companies to break into the marketplace.
Eikka
2.5 / 5 (2) Jul 22, 2012
The manufacturing cost per chip are peanuts.


Because they are made like peanuts: by the millions.

If a chip fab costs $5 bln to build, you're looking at least the same number in research and development. It's on the order of the GDP of a small nation. Suppose the total cost is $15 bln. In order to sell the processors at $100 a piece, you have to make AND sell at least 150 million units before you break even.

Double the investment costs, and you have to sell as many processors as there are PCs sold worldwide in a year - just to make your investment back.

ROI is another matter.


ROI is the matter.

For a new chip to capture 10% of the market share in a year would be phenomenal, which is why it's quickly becoming economically infeasible to keep the Moore's law going. You just can't sell enough of them fast enough to make back the cost in time so you could build the next better fab to keep up with the curve.
Eikka
1 / 5 (1) Jul 22, 2012
If this were the case there would be relatively few very large companies and it would be practically impossible for new companies to break into the marketplace.


Count how many semiconductor foundries exist that are able to manufacture modern processors with the latest processing techniques. You can pretty much count them by one hand.

Most companies actually outsource the actual manufacturing of the chips. AMD, ARM, nVidia, VIA... etc. none actually make their own chips, they just design them and send the drawings off to be made by these big foundries.
SteveL
3 / 5 (2) Jul 22, 2012
If this were the case there would be relatively few very large companies and it would be practically impossible for new companies to break into the marketplace.


Count how many semiconductor foundries exist that are able to manufacture modern processors with the latest processing techniques. You can pretty much count them by one hand.

Most companies actually outsource the actual manufacturing of the chips. AMD, ARM, nVidia, VIA... etc. none actually make their own chips, they just design them and send the drawings off to be made by these big foundries.
Um, OK. But the point was that ROI isn't the only issue for investors. If it were then only those few markets that provide the greatest return would get capital. All others would fade in significance. The fact that there are thousands of stocks and millions of projects worldwide that get capital investment indicates to me that there is more than just ROI going on.