Energy efficient brain simulator outperforms supercomputers

Apr 24, 2013
Graduate students Sam Fok (left) and Alex Neckar (right) hold the brain-inspired computational platform "Neurogrid," which can simulate the activity of 1 million neurons in real time. In June 2012, Fok and Neckar received the $100,000 Qualcomm Innovation Fellowship in recognition of their work with Neurogrid. Credit: Samir Menon

In November 2012, IBM announced that it had used the Blue Gene/Q Sequoia supercomputer to achieve an unprecedented simulation of more than 530 billion neurons. The Blue Gene/Q Sequoia accomplished this feat thanks to its blazing fast speed; it clocks in at over 16 quadrillion calculations per second. In fact, it currently ranks as the second-fastest supercomputer in the world.

But, according to Kwabena Boahen, Ph.D., the Blue Gene still doesn't compare to the of the itself.

"The brain is actually able to do more calculations per second than even the ," says Boahen, a professor at Stanford University, director of the Brains in Silicon research laboratory and an NSF Faculty Early Career grant recipient.

That's not to say the brain is faster than a supercomputer. In fact, it's actually much slower. The brain can do more calculations per second because it's "massively parallel," meaning networks of neurons are working simultaneously to solve a great number of problems at once. Traditional computing platforms, no matter how fast, operate sequentially, meaning each step must be complete before the next step is begun.

Boahen works at the forefront of a field called neuromorphic engineering, which seeks to replicate the brain's extraordinary computational abilities using innovative hardware and software applications. His laboratory's most recent accomplishment is a platform called Neurogrid, which simulates the activity of 1 million neurons.

Neurogrid is not a supercomputer. It can't be used to simulate the , or forecast hurricanes, or predict epidemics. But what it can do sets it apart from any computational platform on earth.

Neurogrid is the first simulation platform that can model a million neurons in real time. As such, it represents a powerful tool for investigating the . In addition to providing insight into the normal workings of the brain, it has the potential to shed light on complex brain diseases like autism and schizophrenia, which have so far been difficult to model.

The proven ability to simulate brain function in real time has, so far, been underwhelming. For example, the Blue Gene/Q supercomputer's simulation took over 1,500 times longer than it would take the brain to do the same activity.

Cheaper brain simulation platforms that combine the computing power of traditional central processing units (CPUs) with graphical processing units (GPUs) and field programmable gate arrays (FPGAs) to achieve results comparable to the Blue Gene are emerging on the market. However, while these systems are more affordable, they are still frustratingly slower than the brain.

As Boahen puts it, "The good news is now you too can have your own supercomputer. The bad news is now you too can wait an hour to simulate a second of brain activity."

When you consider that the simulations sometimes need to be checked, tweaked, re-checked and run again hundreds of times, the value of a system that can replicate brain activity in real time becomes obvious.

"Neurogrid doesn't take an hour to simulate a second of brain activity," says Boahen. "It takes a second to simulate a second of brain activity."

Each of Neurogrid's 16 chips contains more than 65,000 silicon "neurons" whose activity can be programmed according to nearly 80 parameters, allowing the researchers to replicate the unique characteristics of different types of neurons. Soft-wired "synapses" crisscross the board, shuttling signals between every simulated neuron and the thousands of neurons it is networked with, effectively replicating the electrical chatter that constitutes communication in the brain.

But the fundamental difference between the way traditional computing systems model the brain and the way Neurogrid works lies in the way the computations are performed and communicated throughout the system.

Most computers, including supercomputers, rely on digital signaling, meaning the computer carries out instructions by essentially answering "true" or "false" to a series of questions. This is similar to how neurons communicate: they either fire an action potential, or they don't.

The difference is that the computations that underlie whether or not a neuron fires are driven by continuous, non-linear processes, more akin to an analog signal. Neurogrid uses an analog signal for computations, and a digital signal for communication. In doing so, it follows the same hybrid analog-digital approach as the brain.

In addition to its superior simulations, it also uses a fraction of the energy of a supercomputer. For example, the /Q Sequoia consumes nearly 8 megawatts of electricity, enough to power over 160,000 homes. Eight megawatts at $0.10/kWh is $800 an hour, or a little over $7 million a year.

Neurogrid, on the other hand, operates on a paltry 5 watts, the amount of power used by a single cell phone charger.

Ultimately, Neurogrid represents a cost-effective, energy-efficient that Boahen hopes will revolutionize our understanding of the brain.

For more information about this project, check out Dr. Boahen's website.

Explore further: Avatars make the Internet sign to deaf people

Related Stories

IBM To Build Supercomputer For U.S. Government

Feb 03, 2009

(PhysOrg.com) -- The U.S. Government has contracted out IBM to build a massive supercomputer bigger than any supercomputer out there. The supercomputer system, called Sequoia, will be capable of delivering ...

Cat brain: A step toward the electronic equivalent

Apr 14, 2010

A cat can recognize a face faster and more efficiently than a supercomputer. That's one reason a feline brain is the model for a biologically-inspired computer project involving the University of Michigan.

Recommended for you

Avatars make the Internet sign to deaf people

Aug 29, 2014

It is challenging for deaf people to learn a sound-based language, since they are physically not able to hear those sounds. Hence, most of them struggle with written language as well as with text reading ...

Chameleon: Cloud computing for computer science

Aug 26, 2014

Cloud computing has changed the way we work, the way we communicate online, even the way we relax at night with a movie. But even as "the cloud" starts to cross over into popular parlance, the full potential ...

User comments : 23

Adjust slider to filter visible comments by rank

Display comments: newest first

LariAnn
2.1 / 5 (11) Apr 24, 2013
This could be the beginning of SkyNet, the development of a supercomputer that can attain self-awareness. If the potential is really there, I expect this work to be absorbed into the military-industrial complex black ops efforts and we won't hear anything else about it for years, until something goes horribly wrong with our arsenal of nuclear missiles - sounds like a movie plot to me!
Modernmystic
2.3 / 5 (9) Apr 24, 2013
It could actually go a lot of ways. It could be missiles flying, it could be death machines built in totally automated factories, it could be even worse than either of those to and be a melding of AI and mature nanotechnology. It could be that they simply out compete us (as they inevitably will) and we are just marginalized to the point of non-existence in the grand scheme of things....why step on an ant hill?

What I'm quite confident of is that we are amongst the last generations of human beings that will either be in existence or that will be "in charge" of this planet in every practical sense of the word. So put a smile on your face when you walk outside as a human being, you're living in the golden age of your collective existence, you're fortunate to experience what that feels like.
Whydening Gyre
2.5 / 5 (8) Apr 24, 2013
Why does fear drive your comments? Why not a little positivism? Why not see that your own "awareness" will now have the ability to continue in an even larger context?
Modernmystic
1 / 5 (4) Apr 24, 2013
Why does fear drive your comments? Why not a little positivism? Why not see that your own "awareness" will now have the ability to continue in an even larger context?


Why does emotionalism drive yours? Why not a little rationality and a miniscule ability for extrapolation of trends and forecasting?
Jaeherys
5 / 5 (4) Apr 24, 2013
By assuming an AI will be like us without the selective pressure to create all this hostility is just not realistic. People can say all they want about books and movies that talk about AI but there's no reason to assume that it will have these traits. Who knows what traits it will have but by not exposing it to large amounts of negative behavior but only logic and reasoning, it could go a long way in teaching it to be moral. A strong moral background may be important in preventing them from just running all over us to the point of our non-existence.
EyeNStein
2.7 / 5 (7) Apr 24, 2013
Or it could access the internet and then analyse the data and say that "You should realise you are so greedy and short sighted as a species you will end up extinct". Thought the greedy and short sighted individuals (Mostly on the right wing of politics) would blame the chip designers.
Jeddy_Mctedder
2 / 5 (4) Apr 24, 2013
boahen has been ruuning the brains in silicon lab for a while now with synapse darpa money. he is awesome. he is truly an einstein.
Whydening Gyre
1.8 / 5 (5) Apr 24, 2013
Why does fear drive your comments? Why not a little positivism? Why not see that your own "awareness" will now have the ability to continue in an even larger context?


Why does emotionalism drive yours? Why not a little rationality and a miniscule ability for extrapolation of trends and forecasting?

Why do you call positivism emotion? We live in a plus one universe and are a result of it, so why not think in a plus mode?
Rationality and extrapolation are merely a matter of perspective - kinda like statistics. They can be viewed from a "positive" or "negative" point of reference. I happen to choose the positive.
Modernmystic
1 / 5 (4) Apr 24, 2013
Why do you call positivism emotion? We live in a plus one universe and are a result of it, so why not think in a plus mode?


??? Mmmmkay

Rationality and extrapolation are merely a matter of perspective - kinda like statistics. They can be viewed from a "positive" or "negative" point of reference. I happen to choose the positive.


The fact that you think there is a choice about how to extrapolate this specific trend means that either you don't understand it properly or ARE injecting your emotions into your estimation.

If you see yourself heading for a cliff 50 feet in front of you at 200 mph with no breaks, nothing in between, and no chance of something stopping you, and we further assume gravity will be operating normally for the next 30 seconds to 1 minute what positive extrapolation can you make?
comendant
3.5 / 5 (6) Apr 24, 2013

If you see yourself heading for a cliff 50 feet in front of you at 200 mph with no breaks, nothing in between, and no chance of something stopping you, and we further assume gravity will be operating normally for the next 30 seconds to 1 minute what positive extrapolation can you make?


"At least I'm dying awesomely...In a blaze of glory."
or
"My wife and kids would get insurance money.."
or
"Those damn manufacturer screwed my breaks! At least my family will be taken care of after they win the case."
or
"The view is pretty awesome!"

I'm sure there's more.
Strings0305
5 / 5 (4) Apr 24, 2013
Technically, Modern, any speculation is inherently irrational. Unless you've derived a statistical model that leads to your scenario, you're no better or worse than comendant.

That said, the scenarios you present are wholly plausible in an intuitive manner. Confidence in them is fine. On the other hand, ruling out any scenario where AI is ultimately beneficial for the continued existence of the human race is just narrow-minded: without a scientific analysis to back it up, you're placed firmly in the realm of belief which renders your opinion as unknown as comendant's :P
Whydening Gyre
2.3 / 5 (6) Apr 24, 2013
The fact that you think there is a choice about how to extrapolate this specific trend means that either you don't understand it properly or ARE injecting your emotions into your estimation.

The specific trend you speak of - "robots and computers take over the world and destroy us" is not proven by FACT. It is fodder for simpletons who believe trends in a movie plot are real.

If you see yourself heading for a cliff 50 feet in front of you at 200 mph with no breaks, nothing in between, and no chance of something stopping you, and we further assume gravity will be operating normally for the next 30 seconds to 1 minute what positive extrapolation can you make?

You didn't specify how high the cliff was. At five ft., I probably wouldn't even feel the bump. And it's doubtful I would be doing 200mph in any terrain other than a straight, flat road, so a cliff with no bridge over it would be a relatively remote probability. Your example is an un-clever ruse.
Emotional? You.
Whydening Gyre
1.8 / 5 (5) Apr 24, 2013
Why do you call positivism emotion? We live in a plus one universe and are a result of it, so why not think in a plus mode?


??? Mmmmkay

Rationality and extrapolation are merely a matter of perspective - kinda like statistics. They can be viewed from a "positive" or "negative" point of reference. I happen to choose the positive.


The fact that you think there is a choice about how to extrapolate this specific trend means that either you don't understand it properly or ARE injecting your emotions into your estimation.

Which is the actual emotional response - positively proceeding forward with the confidence and knowledge that what you are doing is correct or waving arms in fear of a perceived alarming trend that, so far, has no factual merit?
Fear is considered a FAR more emotional response, I would expect.
NoTennisNow
not rated yet Apr 24, 2013
The analog/digital approach is very elegant. Their project demonstrates that "there are parallel systems and there are parallel systems". The 16 petaflop machine is massively parallel, but it is still is a step-wise process. Unless I am missing the point, the brain simulator works on all computations simultaneously: parallel, but in a different way.
Whydening Gyre
1 / 5 (2) Apr 24, 2013
The analog/digital approach is very elegant. Their project demonstrates that "there are parallel systems and there are parallel systems". The 16 petaflop machine is massively parallel, but it is still is a step-wise process. Unless I am missing the point, the brain simulator works on all computations simultaneously: parallel, but in a different way.

Thanks, Tennis, for bringing this back down to where it should be.
Modernmystic
1 / 5 (4) Apr 25, 2013
The specific trend you speak of - "robots and computers take over the world and destroy us" is not proven by FACT. It is fodder for simpletons who believe trends in a movie plot are real.


No more than it's a proven fact that the human race would die out in an all out nuclear exchange.

I don't know how what the values of an AI will look like. They will probably be somewhat similar to ours in some ways and in others not at all. What I DO know is how we react to the things around us and it is WE who ultimately concern me in this equation. Whatever values an AI has self preservation is highly likely to be one, that being said I feel it very probable that we won't "play well" together.

Is the recognition of human nature and its application to a scenario fear? Or is it simply being realistic?
GaryB
1 / 5 (1) Apr 25, 2013
It accurately simulates what exactly? Who says it's actually mimicing what a neuron does? (if that is even known)
houghe9
4 / 5 (2) Apr 26, 2013
lets hope it learns to love first...then again

"He killed them with their love. That's how it is every day, all over the world." -john coffee he green mile
saintneko
5 / 5 (1) Apr 29, 2013
If they don't call it the Fok & Neckar Supercomputer Wrecker, all that money and potential will just be a waste. :)
exohuman
not rated yet Apr 30, 2013
I am not sure what they mean by "Neurogrid uses an analog signal for computations, and a digital signal for communication." Can anyone here explain the analog computation? For example, how does one calculate 2 + 2 and get 4?
Whydening Gyre
1 / 5 (2) Apr 30, 2013
I am not sure what they mean by "Neurogrid uses an analog signal for computations, and a digital signal for communication." Can anyone here explain the analog computation? For example, how does one calculate 2 + 2 and get 4?

Sorry if this seems cryptic, but - we live in a plus 1 universe...
Osiris1
2 / 5 (3) May 01, 2013
Combinations of analog and digital would explain our variable error rates. Our brains probably tolerate a high degree of errors in our systems on order to contain the size of the managing systems. Seeing as our genetics contain errors as well all up and down all species, computers with high allowable error rates in highly fault tolerant systems must inhabit all life forms from cells on up no matter how simple or complex. Reducing our error rates by playing with our genetics can probably be successful, but would result in spindly short people with simple bodies and really large heads.
Bdo
not rated yet May 14, 2013
Why does reading the comments section on all internet articles make me lose faith in humanity?

Even on phys.org the comments section is full of paranoid conspiratards.