Feeds

back to article HP's Memristor tech - better than flash?

HP will claim today to have pushed Memristor technology to equal the switching speed and endurance shown by current NAND flash cells. The Memristor or memory resistor is said to be a fundamental electrical circuit element, along with the resistor, capacitor and inductor. Its electrical state remains unaltered between a device …

COMMENTS

This topic is closed for new posts.
Coat

No comment.

" For now the best known way to create a human brain is to start with two humans, one man and one woman, and bring them into conjunction. "

I will withold comment on this until I have repeated the expirement myself.

That's right, the one with my name on the label, in my mum's handwriting.

0
0
Unhappy

That's one experiment you don't want to go wrong

http://www.bbc.co.uk/programmes/b00rznhw

Yes, that's my mate Simon playing with his train set. Alone. No he's not really that sad. It's making a point, mkay?

0
0
Bronze badge

"fundamental electrical circuit element"

"The Memristor or memory resistor is said to be a fundamental electrical circuit element, along with the resistor, capacitor and inductor".

That bit just sounds liike hype - resistors, capacitors and inductors are passive devices, the behaviour of which can be simply described in terms of the rate of change of current with respect to voltage. Indeed "pure" resistors and capacitors/inductors are just representations of the two parts of electrical impedance (resistive and reactive). Of course real components always exhibit a mix. However, there simply isn't room for a "fundamental component" to go along side these passive components in the same sense.

There are, of course, many components that change their impedance characteristics with all sort of other factors - with temperature, with the pressence of magnetic fields, vibration and so on. But they are not "fundamental components" alongside resistors and the like.

What this would appear to be is some form of device that can change its resistance characteristics on a persistent basis through an electrical signal. That very, very useful - but it's not a component exhibiting some form of brand new electrical behaviour whcih doesn't fit the standard impedance model.

0
0
Boffin

Brush up on your physics history

"That bit just sounds liike hype - resistors, capacitors and inductors are passive devices, the behaviour of which can be simply described in terms of the rate of change of current with respect to voltage."

Sure, resistance, capacitance, and inductance are the three most well known of the four passive electrical component properties, because their effects dominate at the physical scale where we normally fabricate practical electronic components. OTOH, folks that work with nanoscale electronics are well aware of the hysteretic effects of memresistance.

If you took any graduate-level electronics physics classes, you'll recall that in 1971 Leon Chua published an well-known paper in which he reasoned from symmetry arguments that there should be a fourth fundamental element, which he called a memristor (short for memory resistor). However, although he showed that such an element has many interesting and valuable circuit properties, until 2008 no one came up with either a useful physical model or an example of a memristor, mostly because nanoscale electronics technology wasn't far enough along.

It was Stan Williams' paper in 2008 that showed, using a simple analytical example, that memristance arises naturally in nanoscale systems in which solid-state electronic and ionic transport are coupled under an external bias voltage. These are the results that serve as the foundation for understanding a wide range of hysteretic current–voltage behaviour observed in many nanoscale electronic devices, and for the practical device development being done by Stan at HP.

3
0
Bronze badge

so right.

I suffer from hysterical current-voltage behaviour CONSTANTLY.

DOCTOR! DOCTOR! My brain hurts...

0
0
Pint

It is fundamental because it models a physical phenomenon elegantly.

Memristor is fundamental in the context of describing the phenomenon observed by the HP researchers. Without this mathematical construct, it will be cumbersome to model memristor with a combination of capacitor and voltage-controlled voltage source. Chua's definition simplified the model for the HP researchers and freed their minds so that they can find useful applications for this newly discovered device.

Perhaps this is a confusion between "fundamental" and "linear time invariant". http://en.wikipedia.org/wiki/Passivity_(engineering)

It is true that memristor is not linear time invariant. However, being a device that does not produce energy, it qualifies as a passive device. Nonlinear devices always open doors for new applications, just like in optics.

0
0
Terminator

Someone's been watching too much....

CAPRICA and BSG!

now someone hurry up and invent VR bands...

1
0
Boffin

@Steven Jones

The "resistive" and "reactive" pairing only works if your components respond linearly to voltage and current - in other words, it's just an approximation that mostly works. Drive an inductor into saturation though, and you'll quickly find how much of an approximation it is.

I'm not an expert on this, but the point of this being a "fundamental circuit element" is that you can always calculate voltage from current and vice versa, with no dependence on any external factors. However unlike a resistor where measuring a voltage at one instant will give you the current at that instant, for a memristor you also need to know how voltage/current changed in the past. This is very different from circuit elements such as light-dependent resistors, where their varying behaviour depends on what the outside world does to them.

1
0
Boffin

Nonintuitive, but correct

Chua and other physicists were able to predict the existence of memresistance from basic laws of physics, although they didn't have the technology to fabricate devices small enough for it to be measurable. 40 years on, our ability to fabricate structures at the nanometre level allows us to not only measure and model memresistance, but to build practical, useful components that make use of its novel characteristics.

[Incidently, memresistors aren't the only interesting electronic components that existed in theory but were unrealisable before advancements in nanoelectronics: junctionless nanowire transistors (i.e. Lilienfield transistors) are also likely to develop into useful devices.]

0
0

(title)

"""...for a memristor you also need to know how voltage/current changed in the past."""

That's a big way of saying 'integral,' and in that respect, you need to know about the 'past' to determine anything about a capacitor or inductor, which is what makes them so much more fun than resistors.

0
0

Oskar Heil and the FET

Another example of theory pre-dating practice by years was the field effect transistor which was patented by Oskar Heil in the 1930's.

Chua a physicist? Then how come he was in the EECS department in the College of Engineering at Cal and not the Physics department in the College of Letters and Science? FWIW, he was the prof for my intro to active circuits class (EECS 105) where I was first introduced to SPICE - the input deck was a deck of punched cards.

0
0
Thumb Up

Yeah, Chua's an EE

Yeah, I know that Chua's an EE (I'm personally most familiar with his work on nonlinear circuit theory). But Stan Williams and most of the rest of the team at HP's Quantum Science Research lab have backgrounds in physics (or physical chemistry or chemical physics), not EE. Even my colleagues working on nanoelectronics in my local university's EE department are as likely as not to be theoretical physicists.

0
0
Silver badge

Bah!

Back in the fifties it was telephone exchanges that would mimic the brain. Wire enough of them together in a network structure and they would spontaneously create an AI.

I look forward to the next nitwit statement by an over-caffeine-stimulated latter day lab tech turned technical writer.

At least no-one is "pulling a pluto" and trying to rename the transistor something else.

Maybe something can be salvaged though. How long will it be before Captain Cyborg tries to get the newspapers interested in a project to insert a few memristors inserted into his head?

0
0
Gold badge
Terminator

I think you've just explained BT!

That's a large number of telephone exchanges wired together. It's become sentient and it hates us.

0
0
Gold badge
Go

Might even work better when it gets smaller.

Just a superficial reading of the posts.

But it's got a lot of ground to catch up behind the other technologies.

0
0
Go

But, wow, a lot of progress in 2 years

From reading the paper in the 8 April issue of Nature (‘Memristive’ switches enable ‘stateful’ logic operations via material implication), it's clear that HP has made a lot of progress toward developing practical large-scale components based on the Memristor. (This is a short article readily accessible to the technical layperson.) Another fascinating paper on the subject is Chua's 2003 invited paper (Leon O. Chua, "Nonlinear Circuit Foundations for Nanodevices, Part I: The Four-Element Torus," Proc. IEEE 91, 1830-1859 (2003)).

0
0
Bronze badge

Not the Brain...

I wonder if memristors - or phase change memory - could be applied to FPGA devices. Right now, except for one vendor's "antifuse" technology, the elements in an FPGA that allow signals to be rerouted also cause additional delays.

0
0
Stop

Hardware is one thing

...but how do they propose to program anything but simple storage/retrieval operations, in regards to the uber-memory-device? It's just their quaint attempt at sprinkling some marketroid pixie dust onto the thing, to claim as if it could lead of itself to anything but larger storage capacity.

0
0
Gold badge

it's a new component!

"...but how do they propose to program anything but simple storage/retrieval operations, in regards to the uber-memory-device?"

Umm, how do you program anything at all into a silicon diode? Or a flip-flop? You don't, they are part of a larger circuit. This is a fundamentally new electrical component, I think high-capacity memory will just be the first and easiest thing it can be used for. It should be HIGHLY HIGHLY useful for other uses (not just digital circuitry but analog as well), but this will surely require some new training for electrical engineers to know when and how it makes sense to use memristors in circuits (and changes to EE software as well).

0
0
This topic is closed for new posts.