Nano-sized Microchips? HP Says So.

ImaLamer writes: "A C|Net News story is reporting that HP has announced they have made breakthroughs that 'help turn out powerful computers that fit on the head of a pin with room to spare.' Also in the article, that the patent announced Wednesday, will produce no two chips that are the same. 'Each one will be customized for a particular function,' says Stanley Williams, the chemist on the team. The work was done by himself, Phil Kuekes, a computer architect, and James Heath, a UCLA professor. The chips use nanowires and the chips are said to be even less than the size of bacterium. Sounds cool enough. The biggest part of the breakthrough isn't the chips themselves, but that HP plans to be able to 'fix' chips which come out with imperfections, thus saving money on an already cheap process."

Comments

Interesting story...

[Uttles]

So much so that I posted it this morning, only from the Yahoo! site: HP Says Atom-Sized Computer Chips a Lot Closer

The fact that they are going to be able to fix the chips is a big breakthrough, but the biggest thing here is the process for making the chips. They are breaking the chips into different functional areas, and this is what enables (indirectly) the capability to do "chip fixing."

Re:Interesting story...

[dhovis]

Stan Williams, one of the guys mentioned here, came to my department to give a talk about the work they were doing at HP about 2 years ago.

What they are doing is really facinating, and it's not quite as simple as just re-programming the chip when they come off the line. The chips will continue to develop defects, even during service.

The way they get around this is to design a fault tolerant processing scheme. When you drop the sizes down as much as these people are, you get a several order of magnitude increase in the number of transistors, so you can afford to have the chip do the same calculation, say 500 times in different sections of the chip. The chip itself can figure out what sections are bad, and stop using them on its own.

HP actually built a full size computer where they designed some ASICs that computed using lookup tables (!). They had them fabed and asked the fab to send them the defective chips along with the good ones. They then mixed the good chips and bad chips together (I think it was like a 1/2 good/bad ratio) and hired a high school student to hook up the wiring. Now keep in mind that even on the "defective" chips, part of the chip still worked. It only takes 1 defect to spoil a traditional chip. On the whole, the components on the chips had about a 3% defect rate.

The whole thing ran at a whopping 1MHz and may not have been wired up exactly to specifications, but it was "programmed" with a standard computer first to find the defects and route around them. Performance wise, it was on par with the fastest HP workstations of the day. (there's the MHz myth for you)

So the idea here is to design chips that have so many circuits that you can afford to build in fault tolerance. What is more, you can afford to have the chips constantly checking themselves looking for new faults.

In short, zero defect tolerence is not necessarily a good thing. One defect in one transistor can render a Pentium processor worthless. The smaller you make them, and the more transistors you add, the harder it will be to achieve defect free parts. Yields go down, price goes up.

And if you don't believe me, they published an article in Science about the computer they built (it was called Teramac IIRC)

Re:Interesting story...

[4of12]

No one would by it because they could never guarentee that your PC is going to act the same as everyone else's.

A valid concern, and certainly one that I would have.

Upon further reflection, though, I thought of this analogy:

The brains and nervous systems of any two human beings are absolutely different. Yet, you can program them (education) so that they can perform the same function (eg, produce consistently spelled words of a language.

Of course, programming humans is more involved than programming silicon, but at least it suggests to me that different underlying physical architecture does not preclude having consistent functionality. [Yes, you can argue that the yield of properly functioning humans is not all that great, but, hey, there's hope.]

nanoscience news site

[Adrian Voinea]

If you're interested in nanoscience generally, like I am, or in nano-sized microchips especially, you can find some cool info and news at the nanoscience.ch site.

Re:Neat, but...

[ImaLamer]

The article states that they could be woven into your clothes, yes.

They currently are producing, in some way, these chips. At least enough to test them.

I don't think though, that they will be used as "cpu's" like you maybe thinking. Think devices, medicine, etc.

It would be cool if you had them controlling stuff like your hard drive, and other periph's.

Add in a PCI card Cluster!

CNN reports as well.

[jwachter]

CNN is reporting the story too.

Jonathan

Re:vaporware

[Kaa]

Can you imagine what would happen if this technology were used to manufacture destructive little nanobots that couldn't be seen, but could be inhaled?

Yes.

Moreover, people with a much better imagination and command of language than I already imagined this:

Neal Stephenson "The Diamond Age".

Nanotech in Scientif American...

[GdoL]

An Scientific American article sthat is valued lecture by K. Eric Drexler on "Machine-Phase Nanotechnology: A molecular nanotechnology pioneer predicts that the tiniest robots will revolutionize manufacturing and transform society".

Here you've a story that is a sample of Sci.Am. coverege:

"Purdue University physicist Albert Chang and colleagues have successfully linked two so-called quantum dots such that the tiny structures could conceivably serve as qubits-switches for quantum computers that can be on, off or in a combination of states."

Also you can see more about nanotech here

Here you can see a report on what we can learn from nature when building small.

(When I proposed a similar story...in November it was rejected, because(??) it was basead on a Scientific American)

Teramac

[cweber]

This project seems to be a follow on to the original Teramac project, in which they linked 864 faulty processors together to form a functional and powerful computer. See here.

The real breakthrough then was coping with the defects of the processors and making the whole thing function reliably. It can even detect new faults and route around them (literally). The authors of the paper, chief among them Phil Kuekes, stated back then that this was fundamental technology for eventual molecular computers, which by their very nature would be made of faulty parts.

Now the molecular chips are 'real', and as anticipated, no two of these nanochips are the same. We'll have to rethink our assumptions about machines, QA and such, and take a clue from biology where everything is less than perfect, but can funtion perfectly nonetheless.

HP's recent press release

[WillWare]

Here is a press release, which says in part:

Hewlett-Packard Company and UCLA today announced they have received a U.S. patent for technology that could make it possible to build very complex logic chips -- simply and inexpensively -- at the molecular scale...
[In an earlier related experiment] researchers from the collaboration crossed wires the size of those used in today's computer chips and sandwiched them around a one-molecule thick layer of electrically switchable molecules called rotaxanes. Simple logic gates were then created electronically by downloading signals to molecules trapped between the crosswires...
Once a basic grid has been assembled, programming could be used to implement a very complex logic design by electronically setting the appropriate configuration switches in the molecular-scale structure...
The problem is that on a single large grid all the electrical signals would interfere with each other... The solution proposed by the patented invention is to cut the wires into smaller lengths by turning some "intersections" into insulators... The insulators are created by "cutter wires," which are chemically distinct from the others. A voltage difference between the cutter wire and the target wire creates the insulator.

This addresses what I had seen as the major difficulty to building real circuits out of molecules -- it has gotten easy to build a large regular array of molecular switching elements all wired together in parallel, which is no more useful than a house whose lightswitches are all wired in parallel. I hadn't seen how they'd get the kind of irregular specific wiring that makes useful circuitry possible. This appears to be the answer, or close to the answer.