IBM unveils a chip in neuronal architecture: one million neurons and 256 million synapses - with a current consumption of only 70 milliwatts
IBM unveils a chip that combines neurons and digital synapses, offering - for the first time - 256 million programmable synapses. The chip, which contains 5.4 billion transistors, is the first in the world built on a mass-producible scale - and one of the largest CMOS chips ever built. Despite this high performance, the chip consumes only 70 milliwatts of electricity in full operation and in real-time calculations, figures that are several orders of magnitude lower than conventional microprocessors. IBM's technological breakthrough is presented in an article published this week in the scientific newspaper Science, in collaboration with Cornell Tech - and is a significant part of the way to introduce computers with cognitive capabilities.
A comparison between the processing capabilities and energy consumption of the human brain and those of conventional computers is not flattering to the technology on which the world of computing is based. In order to bridge the performance gaps, the scientists at IBM built a system that did not exist until now: a completely new computer architecture that draws inspiration from the world of neuroscience, and offers a scaling range and a significantly higher level of efficiency than the architecture devised by von Neumann in 1946, which until now served as the basis for the world of computing .
The new chips, the second generation of the research and development project that began almost ten years ago, combine principles presented in the prototype for a single processing core that IBM revealed in 2011, along with a completely new software environment, and a new programming language that was revealed, along with a simulator of the chip, in 2013 .
The new cognitive chip includes a two-dimensional network of 4,096 neuro-synaptic processing cores, each of which combines memory, processing power and communication capability - and operates in a defined event-driven format, parallel, and with fail-safe capabilities. In order to allow the expansion of the system beyond the limits of the single chip, it is possible to attach additional chips that are transparently related to each other, in order to build an infrastructure for future supercomputers based on neuro-synaptas.
Here, IBM is laying the foundations for a completely new world of computers that draw inspiration from the structure of the brain and its operation, with the help of a completely new architecture, scale and expansion range that were previously unknown, efficiency that has no known equivalent in terms of current consumption and processing speed, and the improvement of design technologies at the chip level. IBM estimates that the new chips will make it possible to build new generations of information systems - and their introduction is a turning point in the history of the computing world.
The Advanced Defense Research Agency (DARPA) of the USA has so far invested about 53 million dollars in the project as part of the program "Electronic Systems in Neuron Morphology with Scalable Capability and Plastic Adaptation". Current project partners include Cornell Tech and IniLabs.
The chip itself is manufactured using Samsung's 28 nm technology, which allows for a high density of memory on the chip and the integration of transistors characterized by a particularly low level of current leakage.
The design of the event-driven circuits integrated in the new chips is based on an asynchronous design methodology developed at Cornell Tech, and refined with IBM since 2008. Combining advanced processes and technologies, together with the new and unique architecture, led to a high density and a current consumption of 20 milliwatts per square centimeter – Almost four orders of magnitude (one part ten thousand) less than microprocessors currently known in the industry.
The new chip is based on a comprehensive ecological environment, which includes a simulator, programming concept, algorithms and applications, prototype design models, specification of the neuron components, and instructions in the field of supercomputing that can be implemented on the new chips. This comprehensive ecological environment supports all aspects of the programming life cycle, from the early planning and design phase through development to debugging, installing the application and running it. In order to promote this new technology, IBM designed a comprehensive curriculum designed for universities, customers, business partners and company employees.
IBM's new environment reveals a move designed to bring the processing processes closer to the data itself. Thus, for example, it will be possible to take in huge amounts of data from sensors, analyze them and integrate real-time information in a context-dependent format, and deal with the lack of clarity that often characterizes complex systems that operate in real-time in the real world. The system is suitable for dealing with a wide range of problems, from computer vision, through the integration of data from a large number of sensors and sources, to the integration of intelligence that draws inspiration from the way the human brain works, the internal devices operating within the constraints of current consumption, amount of data and speed.
Looking to the future, IBM is currently building supercomputers by combining a large number of chips on one board, and creating systems that can grow to volumes of a hundred billion synapses or more.
The new architecture is designed to take advantage of future technological capabilities, such as a three-dimensional chip structure and innovative logics that will make it possible to further reduce current consumption, compress components into a denser package, and display higher speeds.
The silicon equivalent of the brain
IBM's new neuro-synaptic processing chips represent a sharp departure from traditional concepts of designing and building computer systems. The new chips create an electronic equivalent to the operation of neurons and synapses in biological systems such as the brain. This comparison is made possible thanks to the use of sophisticated algorithms and silicon circuits with a new design.
Cognitive computers, which will be built around these chips, will not be programmed in the same traditional way known today in the world of information systems. Instead of such programming, cognitive computers are expected to learn from experience, identify connections and correlations, present hypotheses, remember to learn and draw lessons from the results. In fact, this process mimics the way the human brain works, and the learning processes we are familiar with.
In order to meet these tasks, IBM's researchers and developers combine principles from the world of nanotechnology, brain research and supercomputing.
The SyNAPSE project is designed to build a system capable of not only analyzing complex information received from a large number of sensor units and external sources - but also dynamically organizing its internal wiring routes, based on interaction with the environment, in processes parallel to those in which the brain manages to learn and change behavior patterns using a level low energy.
The current project goes far beyond the concept outlined by von Neumann, which has dominated the computing world for the last fifty years. Future computing applications are expected to require functional flexibility that is not offered in traditional systems familiar to us. The new chips are an important step in the development of computers from calculation systems to learning systems. They herald the beginning of a new generation of computers - and new applications of these computers in the world of business, science and government.
Although the new chips do not incorporate biological components, IBM's cognitive processors use silicon circuits that draw inspiration from the world of brain research. Thus, it is possible to build what IBM calls a "neuro-synaptic core". This includes an integrated memory corresponding to that of the synapses in the human brain, together with a processing capacity corresponding to that of the neurons, and communication corresponding to that of the axons in the brain.
IBM's cognitive computing architecture relies on a network of cores operating on a single chip, in order to create one integrated system of hardware and software. This architecture constitutes a fundamental change from traditional computing concepts, which is expected to give birth to more efficient machines in terms of electricity consumption - and without the need for prior programming. The combination between the memory and the processor at the single core level actually mimics the way the brain thinks, which is guided by events and external input, and is carried out in a distributed and parallel process.
The long-term goal that IBM is striving for is to build a chip system that will include ten billion neurons and hundreds of millions of synapses - and consume only a single kilowatt of electricity, and a total volume of less than two liters.
Future chips will be able to process data from complex environments that characterize the real world, using a wide variety of sensors - and act in different ways in order to reach a coordinated and context-dependent response format.
Thus, for example, cognitive systems that monitor climate data for the purpose of long-term water supply planning could analyze the data received from the field regarding temperature, barometric pressure, tidal variables - and predict the changes in precipitation and the needs of the water systems on the one hand, as well as generate tsunami warnings on the other hand. Similarly, traffic lights will be able to monitor traffic data at the intersection in order to adjust their activity cycles to actual loads, and adjust themselves repeatedly with any change in traffic.
34 תגובות
Eyal - Regarding what is written that in the future there will be "a chip system that will include ten billion neurons and hundreds of millions of synapses" you are right that the number of synapses is poor. This is a pen error; It meant hundreds of billions of synapses. It also corresponds to the ratio of synapses to neurons in the current chip.
Eyal
As for the count - the only difference is that Transparent is able to identify up to 10 details, and we are up to 5 (don't get me wrong about accuracy). In both cases - the brain grasps the number without counting.
And about the monkey who learned sign language. Despite what is shown - the monkey did not learn a language - but words in the language. Language is made up of words and grammar. What monkeys taught were words. It's beautiful, but it's not a language. The monkey does not think in sign language.
And so I would say - that these monkeys have additional knowledge, but there is no influence here on consciousness, or self-awareness. I don't think monkeys will know how to solve problems in a better way, or be able to pass information between other monkeys who "know" the language.
"The long-term goal that IBM is striving for is to build a chip system that will include ten billion neurons and hundreds of millions of synapses"
It's a bit strange, because if you compare it to the brain, then the number of synapses should be much greater than the number of neurons (each neuron has an average of 10 thousand synaptic connections with neighboring neurons with which it communicates).
Miracles, you can stop your thoughts for a few moments (it's hard for a longer time...) and just look around, feel the environment. It's a pretty interesting experiment that might give some sense of the way of thinking of a developed non-human primate that doesn't have language.
Regarding counting, I remember there was a study that showed that chimpanzees are able to count to 10 quite easily (there is also the famous experiment in front of the computer screen that you probably know, in which they defeat adult humans) The explanation for this ability of theirs was that when they wander in the forest and meet a rival group they have to count them quickly to know if they should enter the battle or retreat.
I'm also interested in what you said about Helen Keller, and we can ask the opposite question, do chimpanzees who learned sign language (and it's real, I saw a documentary on the subject, it's not an invention) have a higher level of self-awareness than their other group members?
someone
I'm not sure that the size is the determining factor, but inventions of a language mechanism. And maybe you're right, and you really need a minimum brain size for language.
Regarding the language - I will tell you two things. Helen Keller was deaf/mute and completely speechless until her teacher - Ann Sullivan - began to teach her. Helen Keller writes that until her teacher taught her, she "didn't know she existed". She describes a strange world with feelings but no thoughts.
The second thing is - people without language suffer from low intelligence. I mean, we need language to think about certain things. I will give you a simple example. Look at a bookshelf. If there are few books there, say 3-4, you will know how many books there are. Beyond that, you need to count - and for that you probably need a language.
Nissim The question regarding the connection between language and consciousness is a very interesting question and I have thought about it quite a bit in the past, will a human baby who grows up without hearing any human language develop self-awareness like yours and mine? How he will be able to look in the mirror and say to himself without words "It's me!", is an interesting question.
As for the chimpanzee, the size of his brain is a third of the size of our brain, perhaps it is right on the threshold for self-awareness to be formed. Like an airplane that accelerates on the runway but only above a certain speed it will be able to take off, perhaps in the same way only a network of neurons that has passed a certain critical mass can become aware of itself.
Point, according to my understanding, language developed in man's evolution already when he was quite smart.
True, this very evolutionary change helped man a lot to think (he can think in language now) and he also developed a lot following language, but didn't he have awareness before that? If so, then we can and why you call consciousness. For those who can speak. Then it is clear to us why animals do not have consciousness according to your perception.
Eyal
well said. I would expand and say that not all humans have the same consciousness either.
Regarding language in the monkey - I don't think you are right. Humans, and only humans, have unique brain structures designed for language. We know this from a lot of evidence. For example - there are defects in the brain that cause changes in language (I'm not talking about controlling muscles like in stroke victims). For example - we know that different parts of the brain are activated depending on the type of word (noun, verb). For example - various sign languages, which do not require delicate mouth and jaw muscles. It is a fact - that they have not been able to teach monkeys sign language to this day (as opposed to all kinds of nonsense that is read about them, like Koko the gorilla).
Do you need a language for consciousness? If you claim that it is - you also claim that people who are completely without language are without consciousness. That doesn't seem right to me.
Correction - "at an ant or a worm".
point why is it related? Or is it because the anatomical structure of the mouth and pharynx do not allow him to speak? (Similar to the palm of a chimpanzee which, although it looks similar to ours, does not allow it to perform delicate actions like humans, because of the different angle of the thumb)
You treat consciousness as if it were 0 or 1 (existing or not existing) but there are many neuroscientists who think that consciousness is actually like dimming a light in a room, awareness can be weak (as in a chimpanzee according to various tests in front of a mirror) and it can be strong as in us . It can also not exist at all (with an ant it is a worm for example) and it is possible that it can be much higher than ours in an alien for example whose brain is twice as big as ours.
point
Again - you speak without any basis. There is no connection between consciousness and speech. Do you think that a person without language has no consciousness? There are people like that and they have a difficult thinking problem.
If you could explain to us what you mean by the word consciousness - maybe then we can understand you. You throw sentences without foundation and without any explanation. In English there is conscience and self-conscience. The common root in Hebrew is not for nothing.
Please - try to explain what consciousness is, and why you think a monkey has no consciousness. Do you have a test that can tell who has consciousness and who doesn't?
A monkey is not sophisticated enough to even talk, which is an action a 3-year-old child is already capable of doing. So consciousness he must not have.
Father of, consider what I say and not my disabilities simply:)
And there is nothing to apologize for, I did not feel that you insulted me in any way.
I didn't understand which of your senses made you realize that "man is the only one who shows self-awareness". How do you determine this?
post Scriptum.
If you had read the link to the end, you would have seen that certain dolphins, certain birds and certain monkeys did pass the mirror test.
Chipofo's father, Chipofustan
Where do you get a self-aware transparent? Some monkeys recognize themselves in the mirror.
point
If you can't define consciousness then how do you say that consciousness is different from self-awareness?
You said that the color red is in the mind. You said monkeys have no consciousness. So how can a transparent person know what the color red is?
point
And what makes you think that a monkey has no consciousness? I'm asking seriously.
from Oshool
Yes, I rely on sources. My sources are: limbs, eyes, brain, the five senses...
What fact did you not understand? The fact that the most influential animal on the planet is man? Or is the person the only one who shows self-awareness?
And regarding your link to the Wikipedia entry. I didn't understand what you meant by that. It says there that the monkeys did not pass the test. Which reinforces my point.
Besides, you write very well, but I realized that you are dyslexic. So I want to apologize to you, but, I'm not dyslexic. So, sorry.
post Scriptum. Dad, where did this information come from? Is it your opinion or are you relying on sources? If you rely, allow a link please?
Click on the link on my nick for me, normal as you are
from Oshool
You are dyslexic. You must not have seen well.
In other words, period, man understands that he is part of the environment and the monkey does not. Man knows he is man and the monkey is a monkey. But the monkey sees no difference between a person and another monkey. The monkey is affected by the environment (like all other things) but its influence on the environment is small. On the other hand, man is not only influenced by the environment, but also influences the environment more than any other animal. And this fact stems from the fact that man is aware of himself and his needs...
Chipofo's father, to your question, yes.
point
Of course there is a connection. You cannot be aware of yourself without having consciousness. Does "I think means I exist" mean something to you? The person who understands that he understands... he also understands that he differs from the monkey in at least one thing.
And the monkey that reacts to movements in the mirror... so that's exactly what he does: he reacts to the environment. Have you seen an experiment where a monkey responds to movements in a mirror and shows signs that it knows the figure in the mirror is him?
It's hard for me to talk about something that doesn't have a definition. Can you give me a representative example of what consciousness (or a group of things) is?
Like I could give you a list of things that I put in red and tell you to pay attention to their filling, how it looks. Then I can explain to you what red is even without physical knowledge (although it is clear that the color red has a definition that is well defined physically)
Kopipo, there are types of monkeys that react to themselves in front of a mirror. not everyone. In any case, this is an experiment that meets a narrow definition of self-awareness (which is related to the visual perceptual awareness of the production of its own external aspect). If we use a broader definition of self-awareness then it turns out that the definition of self-awareness is realized by a brain. Each mind and its level of self-awareness.
In any case, as confusing as it is, it still has nothing to do with the concept of consciousness.
point
In my opinion, the situation (regarding the monkey) is the opposite. A monkey has consciousness, but it is not aware of itself. (The monkey shows signs of understanding the environment, but at the same time it shows no signs that it, selfishly, knows that it is part of the environment).
Consciousness cannot be defined just like the color red cannot be defined. Red color is part of a conscious experience.
Point, is it possible to know how you define consciousness?
The most trivial example: the monkey that is aware of itself but has no consciousness.
Point, I did see that in Wikipedia there are two different entries "self-awareness" and "consciousness" but the difference between them is so thin that in my opinion for anything and everything they can be used as synonyms.
In any case, it seems that there is a close connection between the two concepts, it is not clear why you claim that there is not.
point
Why do you think there is a difference between the two concepts?
Eyal, despite sharing the words, "self-awareness" should not be confused with consciousness.
At a certain level, a camera that captures the image it outputs through a computer screen is a form of "self-awareness". But there is no connection between this and "consciousness".
Amazing, I wonder how soon these chips will develop self-awareness.
Another step on the way to singularity or transcendence.
A human brain contains 100 thousand times more neurons than this processor. Only in the three-dimensional production of floats will it be possible to pass the barrier of the surface.
Another important thing to note is that the way information is transmitted in the brain is completely different from what this component does.