Scientists Create First Memristor: Missing Fourth Electronic Circuit Element

http://blog.wired.com/gadgets/2008/04/scientists-prov.html

by Alex3917 • 18 years ago

79 40 18 years ago

blog.wired.com

nazgulnarsil• 18 years ago

I don't know enough about electronics to know how sensationalist this article is being.

does it really have all the implications that are mentioned in the article? (analog computing, AI)

gaika3• 18 years ago

Somewhat justified - it allows nano-scale analog processing elements and storage. More details here http://en.wikipedia.org/wiki/Memristor

phaedrus• 18 years ago

It's on the same level as the invention of the transistor.

TrevorJ2• 18 years ago

I'd say it's at least that big.

ashu2• 18 years ago

Some more details here: http://eetimes.com/news/latest/showArticle.jhtml;jsessionid=...

gaika3• 18 years ago

Author is clueless: "changes in voltage, or flux" - in this case flux is not change in voltage, but magnetic field.

ashu2• 18 years ago

http://en.wikipedia.org/wiki/Flux

"Flux is usually the integral of a vector quantity over a finite surface."

So, I don't think the author is clueless. The vector quantity in this case is the electric field.

Edit: although, from reading a little bit more about the Memristor, it appears that it actually relates the magnetic flux to the charge.

  d(phi) = Mdq

So, maybe the author is indeed clueless.

gaika3• 18 years ago

They should have called it flux capacitor, because that's what it is.

Alex39172• 18 years ago

I wonder how many bits will ultimately fit in a single memristor. It seems like this has the ability to increase storage capacity by several orders of magnitude.

TrevorJ2• 18 years ago

This will be big someday, but I fear that the huge inertia that is the software and hardware industry (well, at least the larger companies) will keep this out of mainstream for 5-8 years.

Hexstream2• 18 years ago

Mommy, I can't wait 5-8 years for this utterly game changing technology... I want it NOW! I want it nowWwWWwwWw!!!

rms3• 18 years ago

This should really have a lot more upvotes; this is pretty enormous. Maybe Kurzweil is right after all...

Slashdot comments are decent.

http://hardware.slashdot.org/comments.pl?threshold=3&mod...

Here is the Nature paper: http://www.scribd.com/doc/2764516/nature06932

Alex39172• 18 years ago

The way I'm reading this, our storage capacity today is basically like a bubble wand or a carpenter's level. If you tilt the right side down then the bubble goes all the way up on the left, and if you tilt the left side down then the bubble goes all the way up on the right. This means that bubble wand can hold two values, 0 and 1. With this new process you can essentially tilt the wand a few degrees and have the bubble move only a couple inches and then stop, and then tilt the wand a few more degrees and have the bubble move a little more.

In this way you can store any value within a single "box." So instead of it taking 64 "boxes" to store 64 bits, with this new process we can essentially store this much information in a single box. And furthermore, each box is going to be an order of magnitude smaller than our existing boxes today.

This may well be the single most important day of our lives, technologically at least. The singularity is indeed near.

anamax• 18 years ago

It's possible to store multiple bits in lots of devices, including capacitors and inductors. Intel has been shipping flash that stores multiple bits per cell for quite some time.

There's an interesting tradeoff between the size of the box, the number of bits, and the required detector. And, if you're going small enough, the charge of an electron.

soundsop• 18 years ago

The original paper by Chua: http://www.scribd.com/doc/2753819/MemristorThe-Missing-Circu...

rms3• 18 years ago

This is so big, it's tough to imagine our memristor based future. Please speculate wildly about humanity in the Memristor Age.

I don't think we'll have any keeping up with Moore's Law. In 5 years memristor storage will be everywhere. IBM will develop memristor processors for the Blue Brain project.

icey4• 18 years ago

I need someone with a strong understanding of free markets to help me out here; but if this is really so big, why hasn't HPQ skyrocketed?

I'm not being a smartass, I just really don't understand why we haven't seen financial implications of this find.

Edit In the interest of disclosure, I've bought some shares to put my money where my mind is.

hugh6• 18 years ago

I don't know. Does anyone know if HP managed to patent it before they published it?

icey4• 18 years ago

Okay, I took the time to do some asking around with some finance friends of mine. Their opinions were that:

1) People in that sector don't really understand why this is a big deal.

2) It looks like just another "Breakthrough" in hi-tech; which is to say, a regular every-day occurrence. The fact that all articles point to nothing coming to the market for at least five to ten years also means that finance people aren't interested in the tech yet.

mxh• 18 years ago

"This is so big, it's tough to imagine our memristor based future. Please speculate wildly about humanity in the Memristor Age."

Not since the inductor was discovered has there been such cause for rejoicing.

My guess is that people in this glorious new age will continue to be, by turns, noble, venal, kind, warlike, selfish, generous, stupid, wise, thoughtful and ignorant, and that life will go on much as it always has.

I find it hard to believe that this discovery/invention (as opposed to, say, the transistor, fission, the steam engine, the telegraph, quantum mechanics, et-bloody-cetera) is really going to alter the human experience for all time.

We will, however, have smaller iPods. And this is a cool discovery.

From a narrow, self-interested perspective, I wonder if this will end up having any impact on how computers look to the programmer. I'd guess not; I think this stuff will be used to make faster/smaller versions of what we already have, in much the same way the last 30 years of advances in processor design were used to make really fast IA instruction set chips. If I'm wrong, though, it'll be interesting learning how to program the things.

aswanson3• 18 years ago

Not since the inductor was discovered has there been such cause for rejoicing.

Hilarious. I think that depending on the noise immunity properties they may increase storage density signifcantly, but it will be more along the lines of a sustaining rather than revolutionary product, in terms of impact. I doubt, due to imperfections in the manufacturing process, that we will ever be able to store a continuous analog signal with perfect fidelity. In fact, Shannon's law forbids it has to be sent along a channel, which it does.

Analog computing has been with us forever, we'll see if this makes it commercially viable.

aswanson3• 18 years ago

Not since the inductor was discovered has there been such cause for rejoicing.

Analog computing has been with us forever, we'll see if this makes it commercially viable for some applications. Good news, in any case.

hugh6• 18 years ago

From a narrow, self-interested perspective, I wonder if this will end up having any impact on how computers look to the programmer.

As far as I can see, the most likely case is that you might wind up with a single, huge, fast chunk of memory replacing both your RAM and your hard disk. That should change things somewhat from the programmer's point of view.

mojuba• 18 years ago

I wonder if this will end up having any impact on how computers look to the programmer.

RAM is nothing more than just a cache mechanism for slower storage devices, and as such it could have been made transparent from the programming perspective (roughly the same way as other caching devices - Cache Level 1 ... Cache Level N, CPU registers).

The reasons RAM is still directly programmable are probably historical rather than technical. There is clearly a tendency though in many modern dynamic languages to hide direct memory allocation from the programmer.

So, my first guess is, memristors won't affect much the world of garbage-collected dynamic languages - they are sort of ahead of time already.

There is an interesting perspective, however, if you look at the OS level.

Firstly, some part of the file system which is read-only - mostly code - will not be copied or "loaded" anymore, it will be read directly by the CPU from where it lies in the FS.

Another important change here would be that memory pointers will persist. It is difficult to speculate on the possible impact, but it may be huge. It may even give rise to new programming languages - with or without direct pointer manipulation, I don't know.

As for data files, my guess is that "open", "save" and other common verbs will be eliminated. Instead, we'll work directly on the file and the system will somehow keep the modification history (and here's where the analog nature of memristors might help, or maybe not).

rms3• 18 years ago

Are there opportunities for startups designing memristor based circuits?

Are there any patents on memristors?

newt0311• 18 years ago

the paper was published in the 1970s but the device itself has been fabricate in Hp labs so there is a very good chance that they got a provisional patent on this. They are probably working on a full patent at this time.

jrockway• 18 years ago

Researchers believe the discovery will pave the way for instant-on PCs, more energy-efficient computers, and new analog computers that can process and associate information in a manner similar to that of the human brain.

It will also bring about world peace and time travel.

Hexstream2• 18 years ago

Here's hoping! ;P

Jesin• 18 years ago

Whee! Yet another Wave of the Future!

Hmm. Quantum computers or memresistor-based computers. What to choose? Maybe both, if someone can figure out a way to convert reliably between spin superpositions and resistance.

WTF is all of this going to do to teh intarwobs?

dhs• 18 years ago

For the record: It just occured to me that, with memristors, we would be able to implement large parts of the computational characters of the future directly in hardware. Which would really be swell.

muerdeme• 18 years ago

In fact, you can use any fab to make these things right now

Whoa. I didn't expect to hear that.

aswanson3• 18 years ago

"Building an analog computer in which you don't use 1s and 0s and instead use essentially all shades of gray in between is one of the things we're already working on," says Williams. These computers could do the types of things that digital computers aren't very good at –- like making decisions, determining that one thing is larger than another, or even learning."

How else do they think analog computers worked in the first place?: http://en.wikipedia.org/wiki/Analog_computers

You can implement that stuff with caps, op amps and resistors. I don't know, maybe the substances they use have better differential nonlinearity characteristics across temperature.

All you need is a transistor and the proper bias voltage to create a simple analog computer.

hugh6• 18 years ago

OK, I've finally finished reading the Nature paper. To me the interesting, yet bothersome, thing about this design is that in contrast to the past fifty years of trends in electronics, it actually has moving parts. We're not just sloshing electrons around the place, we're actually moving dopant atoms through an amorphous solid, by applying ridiculously enormous electric fields. My worry is whether this will wind up being durable, or whether the response is going to change after these atoms have been yanked back and forth a few billion times. The paper certainly doesn't say anything about extensively testing these systems.

mleonhard• 18 years ago

This discovery should lead to very cheap NVRAM. What new ways of doing business will arise when a computer with 32x4GHz cores and 16 TB of memory costs $2000 TCO/year?

Disks will no longer be the performance bottleneck of cutting edge systems. Instead it will be network latency. I think that a very important problem will be the geographic distribution of information to minimize latency for users and maintaining consistency of those replicas.

Additionally, the current model of software installation will become obsolete. A very important current problem is how to deliver applications over the network with good performance and stability. I think that we need something better than the browser/DOM/JavaScript and the proprietary offerings from Adobe, Microsoft, and Sun.

These are two important problems that I am studying. What hard problems are you working on?

danohuiginn• 18 years ago

Woah. Positive correlation between how excited people are by a piece of science news and how much they know about the topic. How often does that happen?

Scientists Create First Memristor: Missing Fourth Electronic Circuit Element

Related Stories