Slashdot Mirror
Memristor — 4th Basic Element of Circuits
esocid writes "Researchers at HP Labs have solved a decades-old mystery by proving the existence of a fourth basic element in integrated circuits that could make it possible to develop computers that turn on and off like an electric light. The memristor — short for memory resistor — could make it possible to develop far more energy-efficient computing systems with memories that retain information even after the power is off, so there's no wait for the system to boot up after turning the computer on. It may even be possible to create systems with some of the pattern-matching abilities of the human brain. Leon Chua, a distinguished faculty member at the University of California at Berkeley, initially theorized about and named the element in an academic paper published 37 years ago. Chua argued that the memristor was the fourth fundamental circuit element, along with the resistor, capacitor and inductor, and that it had properties that could not be duplicated by any combination of the other three elements."
Practical use, probably 15 to 20 years.
Insanely expensive prototypes with virtually no functionality in modern use, probably 4 to 10 years.
Full Tilt
One more thing to wipe after surfing porn.
In about 200 years' time, when Evil returns.
As far as I know, human brains don't retain much information when the power is turned off and there's usually some trouble after the power is restored. Furthermore, I'm not sure how power-down information retention relates to pattern-matching abilities.
But what to I know, I had my brain off last night.
It must have been something you assimilated. . . .
Basically you have Ohm's law which is v =Ri. There is a component for each variable: Capacitors for voltage, inductors for current, resistors for resistance. It is all there, in nice little differential equations.
Yes, this is a great discovery. But please stop with the sensationalist headlines. This is getting out of hand.
An engineer, a physicist and a mathematician are staying in a hotel.
The engineer wakes up and smells smoke. He goes out into the hallway and sees a fire, so he fills a trash can from his room with water and douses the fire. He goes back to bed.
Later, the physicist wakes up and smells smoke. He opens his door and sees a fire in the hallway. He walks down the hall to a fire hose and after calculating the flame velocity, distance, water pressure, trajectory, etc. extinguishes the fire with the minimum amount of water and energy needed.
Later, the mathematician wakes up and smells smoke. He goes to the hall, sees the fire and then the fire hose. He thinks for a moment and then exclaims, "Ah, a solution exists!" and then goes back to bed.
Hey Taco! Looks like you're using the "infinite monkeys and typewriters" scheme to generate Ask Slashdots again...
I don't understand what makes it a "fundamental" part of a circuit, while say a diode or MOSFET isn't. You can't make a transistor out of resistors, capacitors, and inductors... That's why it always showed up as the magical "voltage-controlled current source" in entry-level circuit analysis courses. I thought the three classic "basic" elements were because they were just the simplest.
Or maybe they're "basic" because every circuit (that's not superconducting), whether or not it contains semiconductors or more exotic stuff, has some amount resistance, capacitance, and inductance. Even if you don't want it, in which case you call it "parasitic". I don't think you're going to accidentally create two separate layers of titanium oxide.
So while I get why this discovery is totally awesome, I don't get what they mean by "fourth fundamental circuit element". Anyone got the skinny?
The enemies of Democracy are
and then we'll have Leeloo and her multi-pass! Totally cannot wait...
http://technology.newscientist.com/article/dn13812-engineers-find-missing-link-of-electronics.html
This is very interesting stuff. I wonder if these will ever be produced for amateur use, or if they'll only ever find their way into DRAM and such..
- doctea
http://www.eetimes.com/news/latest/showArticle.jhtml?articleID=207403521
- doctea
As for the others, they are components. For instance, a resistor R fits in dv = Rdi. A capacitor C fits in as dq = Cdv. An inductor as dphi = Ldi, and a memristor fills in the missing dphi = Mdq.
We've had Non-volatile state storage for ages (eg. FeRAM and floating gates (as used in flash) and battery backed up RAM). State storage is only part of the picture.
Whatever the mechanism, freezing state is not sufficient to instantly boot a modern computer. Pretty much all modern computers have some communication with an external device that needs to be renegotiated and reconnected, be that a mouse, disk or network.
Engineering is the art of compromise.
Good: You're trying to recognize privacy problems.
Bad: You apparently don't understand the problem well enough to differentiate problems from non-problems.
-- The act of censorship is always worse than whatever is being censored. Always.
This one took quite a bit of thinking, although this wikipedia article summarizes it best.
A transistor may be approximated as a variable current source. Similarly, many applications of transistors are as "active" devices, which supply external power to the circuit being considered.
A diode is effectively nothing more than a voltage-controlled switch. In a DC circuit, it simply passes current through (with a small voltage drop that can be approximated by an inline negative voltage source).
Likewise, all transistors can be abstractly considered as networks of diodes. This is why they are inherently binary devices, and why computers "think" in binary.
The classical circuit elements (Resistor, Capacitor, Inductor) each fundamentally affect the electromagnetic properties of the electrons flowing through said circuit.
Resistors impede the flow of current; a capacitor is a current "bucket" that also blocks DC signals in AC circuits; and an inductor builds up a sort of inertia for the flow of current, through the creation of a magnetic field.
The distinction is hazy, but I think I can see it where it comes from.... when a diode/transistor does something, it affects of the "layout" of the circuit, rather than directly affecting the electrons flowing through it.
The memristor is extremely interesting, as it blurs the line between analogue components and solid-state devices, and provides exciting possibilities for the development of analogue computing and data storage.
Even more exciting is that they can already be made smaller than transistors, and two can be combined to create a device that functions analogous to a transistor.
Considering that we're quickly approaching the limits of Silicon-based technology, this invention may very well offer the key to the true "next generation" of electronic devices, and may very well be as significant to our generation as the transistor was to the previous. This is Nobel Prize-worthy stuff we're talking about.
Kudos to HP for supporting "true" R&D. They most definitely will be reaping the benefits of this one for years to come.
-- If you try to fail and succeed, which have you done? - Uli's moose
I don't get what they mean by "fourth fundamental circuit element"
There are four fundamental circuit variables; current, voltage, charge, and flux.
We can define the relationships between charge and current and between flux and voltage. (charge as an integral of current, flux as an integral of voltage over time)
A resistor provides a function to relate voltage and current.
A capacitor provides a function to relate charge and voltage.
An inductor provides a function to relate flux and current.
Until now we did not know how to construct a passive device which would provide a function relating charge and flux. The only remaining combination of these fundamental variables.
The difference between a memristor and FeRAM is that because the memristor is constructed without using any transistors, it can be used as a kind of analog memory. Instead of just storing 1's and 0's, it's resistance is an analog value anywhere in the range of on and off. Of course you can still use it to store digital data, but the real fun will come when you interconnect these things to emulate the analog behavior of the brain. This is where the claim of pattern recognition and facial recognition come in. They're not actually talking about software there but the actual analog capabilities of circuitry built with memristors.
The other amazing thing about memristors is how small they are. The articles state that you can emulate a transistor by connecting a few memristors, and that transistor is smaller than any we have today. Also it states that the memristor actually performs better at smaller sizes. This really is neat stuff.
The three elements we're used to (R, L and C) relate four things, potential, current, flux and charge. R relates current and potential, L current and flux and C potential and charge. The thing that relates flux to charge is this newfangled (compared to the other three) thing called a memristor. The other two relations (potential/flux and current/charge) are fundamental conservation laws.
At least that's what my quick Googling on the subject turned up.
The real "Libtards" are the Libertarians!
So what you're saying is that it's sorta like a capacitor, but instead of voltage, its function operates on flux.
How many gigawatts can it handle?
thank you Aziz, that's much better
*blink* *blink*
I work for the Department of Redundancy Department.
Exactly! After reading the few articles, wikipedia and the available information from HP, it looks more like a generational change in technology rather than just a new kind of memory. I think the Nature article's wording of discovery is correct here, this looks like an interesting piece of base research with large real world applications, instead of a specific invention to store things.
Given that this memristor looks like to be using very little power, can be scaled down very well and can be used both as storage and to build transistors - I'm pretty excited about this. Yeah, there are other attempts at non-volatile ram, but they are either slow (flash), cannot be written to many times (flash), expensive (a lot of flash alternatives) or just simply too energy consuming, the memristors should bring in some nice competition into the field, since the articles specifically state that it doesn't generate much heat at all, compared to currently existing other technologies, it can be made to change state faster than they could measure(!) in the lab and it can be repeated many times. So, the only part that is left is whether it is economically feasible to mass-produce these. I'm guessing it shouldn't be a very large problem either given the relative simplicity of this discovery.
It takes a man to suffer ignorance and smile
Be yourself no matter what they say
Why do you think it will be such a long time?
IBM discovered GMR and that was nearly universally used in hard drives within ten years.
Closer to 200 atoms wide. Take a glance at http://hyperphysics.phy-astr.gsu.edu/hbase/solids/sili2.html and notice the arrangement of the crystal (face-centered cubic) and the cube size. Going along one face, we pass two atoms before reaching the other side (one corner and one face center), so we have two atoms per 0.5 nm. Now chips are being mass produced on a 45 nm scale. This is about 100 times the silicon crystal cube size, with two atoms linearly per cube.
SIGSEGV caught, terminating
wait... not that kind of sig.
Well, for starters, a transistor has a base, an emitter and a collector. A current flows from the base to the emitter which also lets another current flow from the collector to the emitter.
A memristor only varies the resistivity from one of the wires, which effectively isolates the two circuits. This cannot be effectively be achieved with resistors, capacitors, or inductors. So, the memristor is actually a "resistivity transistor", which happens to have memory included.
A practical application for this would be a digitally-driven analog volume control for your stereo. The + / - buttons would apply a current to change the "virtual knob"'s memory. So the next time you turn on the stereo, it'll have the same volume that it was when you turned it off. No mechanical wear and tear, and no batteries required.
And that six-transistor radio I listened to as a young boy must have been receiving digital signals in 1965.
Look what the computer age has done: analog has been forgotten, even by bright minds.
But luckily not by everyone. The brightest minds still are aware that even the fastest digital 'gates' are fundamentally analog. Actually, especially the fastest....
But repositories of common knowledge are being filled with well-intentioned, but less-than-half-baked treatises and misinformation.
Transistors are IN NO WAY networks of diodes. Yes, some (bipolars) have a p-n junction, but there is no way that diode theory explains amplification. And don't get me started in fets (junction or insulated-gate). No diodes in those (except the packaged-in protection diodes in some). The jfets can be used to rectify, but that's not their nature.
I'm assuming some CS major wrote that wiki about something they didn't understand. Please don't be offended - that's not a stab at CS majors, but it was obviously someone "web-active" and had some (but inadequate) exposure to electronics.
(Disclaimer - I didn't read this wiki, so I'm taking for granted that moosesocks has lifted it from a wiki verbatim)
Ugh.
And to the GP - there is a fundamental difference between passive and active components.
To extend what you said, there are two problems we're seeing with current transistor technology.
Most inexpensive chip processes involve using blue and UV light to effectively (with chemical baths and deposition treatments) etch the surface of the chip into the correct shape and size. The biggest limitation here is that the light is around 300nm. When you're working significantly smaller than that, etching with this light is not as effective. Using higher frequency light can break down the materials (bathing things in xrays is generally a bad idea). One solution is to use refraction, which brings the wavelength of light of a given frequency to a smaller scale without increasing the energy of the light. This helps when etching smaller features (45nm is close to the limit of this hack).
To make smaller structures, they either need to grow them up, which takes time, or they need to build them individually using finer systems like electron guns. This can allow for tests on impossibly small transistors to see their behavior but are inefficient to produce for mass marketing.
The other concern is that with transistors, since you have charge being separated across a barrier (in a FET at least), the smaller the barrier is, the more likely you're going to breach it. If I remember correctly, and I may be wrong here, this should be related to leakage current.
I don't see us being able to make components smaller than the minimum etching size of 45nm. Since a transistor is usually larger than the minimum size (allowing for room for a gate and terminals on both sides, plus the width to allow for contacts to the silicon), this might still be an improvement.
What I'm wondering about is when people are going to realize that static memory prevents the benefits of a reboot. If a system is shutdown and its memory is still flooded by a program with a memory leak, it may not be recoverable. Better operating systems are handling this nicer, but I still need to reboot my machines from time to time to free up a few hundred megs of ram. I can imagine this might completely foobar a few machines without adequate memory cleanup.
Any thoughts?
I wouldn't consider the mad hatter mad. Just reality impaired. He sure can make a mean cup of tea.
I think where you're really going astray is in assuming that a system which doesn't typically need to perform a full reboot would somehow be incapable of doing so; you even imply OS developers might forget that's a requirement, which is a huge insult to every OS developer. Having a system be able to boot itself from a clean slate (as it would have to for a new install, replacement of memristor-RAM sticks etc.) is not only a blindingly obvious requirement, but code that would be required anyway to boot the first system. Of course they'll be able to perform a full reboot, even if doing so is not the default.
As to the comparison with hibernate, it sounds far more like suspend than hibernate. On a system with memristor-RAM everywhere, including the peripherals, suspend would be much simpler - you just stop the clocks and cut the power. You'd still need to code for the apparent huge shifts in external time suspending would produce, but you have to do that now, in addition to saving and restoring the internal state of peripherals and associated data structures.
Chernobyl 'not a wildlife haven' - BBC News
Well, for starters, a transistor has a base, an emitter and a collector. A current flows from the base to the emitter which also lets another current flow from the collector to the emitter.
You have described a bipolar transistor. However, with field effect transistors, there is no current between the gate (base) and the source/drain (collector/emitter). In the case of a bipolar transistor, the emitter/collector current is controlled by the base/emitter current, whereas the source/drain current of a FET is controlled by the gate *voltage*.
The transistors used in CPUs are generally FETs.
http://blog.nexusuk.org