Slashdot Mirror
Cascading Molecules Drive IBM's Smallest Computer
Benoit Fries writes "EE Times reports that IBM researchers have created a simple computation engine that's more than 250,000 times smaller than the most advanced silicon circuitry. Called the world's smallest computer, the system relies on a 'molecular cascade' that pushes a handful of carbon monoxide molecules across a copper surface to perform digital logic functions. 'Even if CMOS density follows Moore's Law for 40 more years, molecular cascades are still going to be smaller,' they said."
I think IBM is going off the wrong direction in tackling Moore's Law.
We should be attempting massive parallelism instead of packing more logic per area.
Isn't that how our brain works?
'Even if CMOS density follows Moore's Law for 40 more years, molecular cascades are still going to be smaller'
Pfft - if I had a nickel for every time I heard that...
Counter Strike for mice.
Carbon monoxide? Carcinogenic hard drives! I was worried about my computer being too safe.
Have you been stalked by Seth today?
So if the power goes out, half the city asphixiates, right? :-)
blog |
*LoC == Standard metric unit of information (Library of Congress). Size of unit varies from year to year.
---- El diablo esta en mis pantalones! Mire, mire!
Niiice. This means we don't have to learn new calculus to program assembly and STILL experience the computing power of single atoms. Good. My head hurts when thinking about sets AND super-sets at the same time (read, quantum computing)
I am the Barber of Seville.
'Even if CMOS density follows Moore's Law for 40 more years, molecular cascades are still going to be smaller,'
Chances are it'll be more than 40 years until they could make an actual product with this technology so I don't think that I'm going to hold off on getting that new conventional cpu quite yet
People seem much brighter once you light them on fire.
Excuse my ignorance, but what are the real life applications of this technology?
I'm guessing medicine, but does anyone have any good ideas on how to use it?
"The slow operation of the gates -- some required seconds to settle -- underscores the fact that the work was part of a research project."
You pretty much have your choice of one chip that does something 250,000 times in a second, or 250,000 chips that do one thing each a second... Until they can speed these things up, they're more of a curiosity than a useful technology.
Nooo!!! You're stepping on them!!!
It's not the size that counts, it's how you use it!
That was so unexpected.....yeah, right.
--
http://nemilar.net - Not your grandmother's soup kitchen
All those people worrying about asphyxiating on carbon monoxide...
CO binds very tightly to metallic ligands such as copper. The Carbon atom has an unbound lone pair of electrons, that are donated to the metal's d-electron shell. Additionally the CO molecule creates a pi-back-bonding system with the metal center, making the complex even more stable.
Upshot: the CO is not going to spontaneously leak off the chip into your atmosphere. In any case, I doubt that such logic circuits would contain sufficient carbon monoxide to pose a health threat.
(Interesting side note: CO asphyxiates you by binding very tightly to the iron in hemoglobin in your blood, much more tightly than oxygen can. IIRC, however, CO will preferentially bind to copper over Fe.)
My other sig is also a
The first person who makes a Star Trek joke about Cascading Failure gets shot.
Aw, I was almost getting excited as I read the article. This technology appears to be a long way from being a post-silicon circuit alternative for CPUs.
:)
It's "exceedingly slow," according to the article. Still, maybe some kind of niche exists for it to be useful. Then again, maybe they'll implement the NOT gate and get this puppy running near the frequency of 500nm light or something.
I'd be excited by that.
250,000 times smaller than the most advanced silicon circuitry. Of course, it's also 250,000,000 times slower. I'm guessing there won't be molecular cascade chips in my PC anytime soon, unless I have a lot of free time...
Because it is not radioactive.
LOS ANGELES 6:39PM PST - The American Assocation of Midgets issued a press release stating "finally a computer company is aligned with our cause. We, the worlds smallest people have been waiting for decades for the worlds smallest computer."
Somebody correct me if I am getting this whole thing wrong, but AFAIK, when you go down to molecular levels, due to the uncertainty principle, sometimes the dominos will not fall as you predict, becauese either
1) they were already fallen you just didn't know, or
2) statistically speaking there is a much higher chance for "spontaneous reverse-thermodynamics" on a molecular level.
what i mean is that while macroscopically speaking, the universe is headed toward higher entropy, molecularly speaking, it's not necessarily so; The example commonly given is that you can drop and shatter an egg, or an shattered egg can come together, absorbing the sound waves etc and rise back into your hand. the latter will not (or, has completely ignorable probability of) happening, but as you and the egg gets smaller, the chance of this ignorable probability becomes less so.
hence, a molecular computer has the probability of operating "faultily" because of the laws of thermodynamics is not followed 100%. this is currently overcome by the thousands / millions of electrons we send over gates, probabilistically speaking they still behave on a macro level, but a molecular computer has no such luxury.
i mean, even there was only a minute chance that one molecule will go backwards as what we intended -- counting up the billions of calculations per second we expect from each chip, and the number of chips out there, and then the number of seconds / days / monthes / years they are expected to operate, the chance of error is almost inevitable. some serious redundancy / self-healing hardware / software might need to be invented.
i am just blabbing, though. like i said: i am no molecular physicist, so if there are some here, please comment.
My life in the land of the rising sun.
Tiny computers were there first, and I believe they even have a patent for the worlds smallest computers. Pictures of the products on their web site are actual size.
What if we're all part of some gigantic computer and the molecules we put to work computing were already computing something ?
Is God going to sue us for stealing processing power ?
graspee
...it still wouldn't be large enough to connect a network cable.
Read The Diamond Age by Neil Stephenson. He talks about rod logic in there, similar to what they are doing at IBM.
Wasn't that what caused all the aliens to pop up in 'Half-Life'?
RMN
~~~
... is that AMD chips run on smoke, and IBM chips run on Carbon Monoxide.
You can read the express paper at Science.
From the article:
The most complex circuit they built is so small that 190 billion could fit atop a standard pencil-top eraser 7mm in diameter.
In my days, when you wanted to show something was really small, you counted how many you could fit on the end of a pin, or in the width of a human hair. Comparing it with something that's almost 1 cm across is cheating.
RMN
~~~
My first thought was, the structure once toppled, IS toppled, and with a stationary background, it would not be possible to reset it. I found it is indeed true. In the IBM page, it states
...It takes several hours to set up the most complicated cascades. Since there is no reset mechanism, these molecule cascades can only perform a calculation once....
My idea is, have a non-stationary background of copper plane, which through some mechanism (which causes repulsion of the CO molecules) places the molecules in the reset position, ready to be "toppled" again!
"Do something man. Right now."
Can it work /twice/?
They compare this to a domino effect. I dont recall any dominos volenteering to set themselves back up. Is this just one-shot proccessing? Nice idea, but I dont think this is the future, not in current form anyway.
-- 'The' Lord and Master Bitman On High, Master Of All
For Americans who are unfamiliar with international units such as an Azerbaijan, it slightly smaller than Maine.
The CIA website provides a convient and fairly comprehensive translation table between US units and international units.
-
- - You can't take something off the Internet! That's like trying to take pee out of a swimming pool.
You have to learn entirely different programming methods to program algorithms to run in parallel. Managing memory and cache access between multiple processors is a pain in the ass on the hardware side. That's what makes mobos for multiple processors more expensive. Plus, some tasks are just not well-suited to scaling across multiple processors at all.
In short, I'd rather have a one processor machine over a two or more processor machine if the one processor machine gives sufficient speed for a reasonable price.
Maybe the state's highest function is to grind out insoluble problems. (Zelazny, Hall of Mirrors)
there is a nail stuck in a piece of stone for 200 years.
the nail has fused itself into the stone.
there is a glass window pane, it has slowly melted
into a warbled surface, so the light passing through
it and coming into my room is no longer uniform.
the smaller you make it,
the less long it will last.
the 0.20 micron chips will last longer
than the nano-chips made 10 years later.
cheers!
john
> What if we're all part of some gigantic computer and the molecules we put to work computing were already computing something ?
I tend to think that we're more likely the mung that's collected inside God's keyboard...
"News Flash! Hubble Telescope Detects Giant Fingernail Clipping and Cluster of Muffin Crumbs"
Cheers,
Jim
-- My Weblog.
If a cascading molecule NOT gate is hard then thier cascading domino metaphor must not be accurate... It's easy to build a domino NOT gate. Here's how:
:
TtttttttttttttR
i
i
I
It's 2 runs in an L shape. Simultaneously gate a True signal at T and the input signal at I, read the result at R. Note: True = Falls, False = Stands.
Here's how it works
If I = True then the shorter I run knocks down the last t. When the longer T run reaches R, the last t will already have fallen so R will not fall. so we have:
I = True --> R = False
If I = False then the T run will knock down R. So we have:
I = False --> R = True
That's a NOT gate!
Combine that with a V shaped OR gate and you have a NOR gate. It's well known that any logic function can be constructed from NOR gates.
Jonathan Weesner
this is it -- this is the way computing will go. ultra-small, using tiny amounts of energy (you could power a tiny supercomputer just with your body heat. a few breakthroughs from now and combined with advanced MEMs, the possibilities are frightening.
You just hold it upside down and shake it.
Let's not stir that bag of worms...
- Simultaneous Multiprocessing, a technology said to allow several hundred instructions to execute through the same physical wires and gates simultaneously. This allows Intel to reduce the transistor count from 948,089,112,552 transistors, as in the Pentium 6, to 14 transistors. (Plans for the next revision include dropping one of the remaining 14 transistors for cost effectiveness.)
- Temporal Result Ordering, which uses a built-in fluxcapacitor to efficiently move instructions and data backwards and forwards in time. This allows the processor to execute code during idle cycles and deliver the results to processes that have already finished executing, or will begin executing at some future time. This provides an incredible boost in speed and efficiency because:
- The processor can use the result of a computation before the computation itself is executed, and even before the program that contains the computation is loaded into memory.
- Computations whose results will be used at some future time can be performed early, before the user even decides to run the program.
- SpiritRun Technology, an extension of Temporal Result Ordering, which allows the processor to execute program code by its spirit, rather than its letter. As all programs contain bugs, or programmer errors which lead to undesired program behavior and crashes, this technology will save businesses over $80 billion dollars per year in lost data, staff time and resources. SpiritRun uses Temporal Result Ordering to detect crashes before they occur (again, during idle cycles taking place in the past, present or future) and analyses the program in its entirety to determine the cause of the undesired operation. At this time, the processor automatically corrects the program code to provide the desired operation. This technology also makes all code 100% secure because the processor detects crackers before they're even born and automatically modifies the holes that allowed them access in the first place.
- Built-in Photorealism Processing Unit, which generates photorealistic graphics by allocating a parallel universe which physically contains a perfect replica of the object being rendered and a photographer. The photographer takes a perfect photograph of the subject and it is digitally transmitted via the Interverse to the processor. Because the parallel universe has a timeline of its own, completely separated from our perception of time, this information appears to arrive immediately, even though the photography may take several hours in the parallel universe.
- Built-in Orchestra Sound Unit, which generates sounds for audio applications which rivals that of the greatest orchestras in the world. This works similarly to the Photorealism Processing Unit, except that a parallel universe is created which contains an orchestra. The sound is recorded and transmitted, again, appearing to arrive immediately, even though the orchestra may have practiced the piece for years in the parallel universe.
As you can clearly see, AMD has a lot of catching up to do.Kind of like when Alexander Fleming wrote up a journal paper back in 1928(?) about how mould killed bacteria, and Walter Florey found it in a literature search a decade later and set his research team to isolate the responsible compound and figure out how to produce it in bulk.
I've had this experience myself. I needed to find an efficient algorithm for a relatively obscure problem. The usual textbooks didn't help, but I finally located a survey paper which finally revealed a 1981 journal article which described exactly the algorithm I was looking for.
Any sufficiently advanced technology is indistinguishable from a rigged demo
--Andy Finkel (J. Klass?)
DNA is an example of a molecule where the position of billions of individual atoms matters. And, guess what, it's quite stable and it works very well for information storage.
Also, their math doesn't make sense.
They say it's 250,000 times smaller than current tech, then they say it's better than current tech plus 40 years of Moore's Law.
Moore's Law states a doubling period of 18 months, or 1.5 years. This gives 26.666... doubling periods for 40 years. So, "if CMOS density follows Moore's Law for 40 more years", it will be 2 ^ 26.666 times smaller, which is in the neighborhood of 106.5 MILLION... that's more than 425 times smaller than "250,000 times smaller".
To reach 250,000 times smaller, under Moore's Law, 27 years would be more than enough.
"Mind, as manifested by the capacity to make choices, is to some extent present in every electron." -Freeman Dyson
It's an interesting project, but that's a long way to go....
"A language that doesn't affect the way you think about programming, is not worth knowing" - Alan Perlis
After all, 4.4585 yAz equals one square micron. It's a much more sensible unit. The sorter is around 900 micro-yAz.
WWJD for a Klondike Bar?
Intel experienced a lot of resistance from engineers when they introduced their first microprocessor. Some of them joked about losing their computer in the cracks between floorboards, although the main problem was that they couldn't get their mind around replacing the whole CPU if one transistor on it failed.
Okay, so these things aren't into speed. They could work great for storage though.
________________________________________ History Must Not Fall Into The Wrong Hands ___________________________________
The fact that the table exists doesn't surprise me that much (in other words no assinine waste of taxpayer money surprises me anymore.) My question is: What job do you have that you would happen to have this link on hand, and how might I apply. (Or did you just figure such a unit conversion must exist and do a google search.) I await and fear any reply.
"Do it in parallel" is one of those great buzzwords in tech that claims to solve all problems, but ends up failing (for what it's worth, "Use a neual net!" is another of my favorites).
.002 fps.
People, the costs of parallelizing a given problem are LARGE. It works best for iterative problems, where you require little inter-processor communication. By splitting your chip in two (effectively), you are reducing the communication between the two parts to that of the front-side bus, which is much slower than within-chip communication. Folding@home works in parallel because the jobs are easily distributed. But many things don't work so well. Imagine a graphics card trying to do this, frame rates would be like
This is why, if I have a certain number of transitors to "play" with, the fastest chip has them all on one die. Parallelizaion is only done when you just can't find a chip fast enough to do what you want to do (Think Beowulf clusters).
And actually, our brain models best, I think, as a single chip, maybe as a 2-chip system. Yes, it has regions devoted to different tasks, but so does a single chip. Also, the fact that our brains have two hemispheres is a severe detriment - this is why we can only control one of our hands particularly well. The reason is the same as with a 2-processor PC - information transfer is slow between the boundary. If we had a one-hemisphere brain, we would be much more capable.
-Looking for a job as a materials chemist or multivariat
> Second of all thank god air molecules aren't actually traveling around the room at 300m/s.
:(
:)
Your god isn't helping you very much, apparently
> I mean a tornado can send a pencil thorugh a telephone pole at a slower velocity.
A pencil masses ~5 grams. An average CO molecule weighs 28.009 atomic mass units. The pencil is 10^23 times heavier. That is a _lot_, and has a big influence on pole penetrating ability.
> Thanks for showing why you can't indiscriminately put two energy equations together and come out with the right answer.
This is true, but E=kT*3/2 does give the average kinetic energy of particles in a gas. Thus, it is not at all indescriminate to use this to find the average velocity.
Air molecules really do go zipping around at hundreds of meters per second. (RMS velocity of air molecules at 15C is ~500m/s.) They are not very massive, so each one doesn't do much. Moreover, at atmospheric pressure, there are so many collisions per second that everything averages out really really well. The chance of a large enough imbalance of air molecules all hitting a pencil from the same direction and accelerating it to tornado velocities is infinitesimal. (Think thermodynamics and entropy: a lot of molecules all going in the same direction would be a lower entropy state than the usual random directions. Starting with a high entropy, random direction, set of air molecules, you're going to have to wait more than the age of the universe to see them all going in the same direction (for a large amount of air molecules, where large is a number at least big enough to make my claim true).)
If you don't believe me (even though I have a physics&CS honours degree), go look it up. I got the 498 m/s RMS speed for molecules of dry air at 1atm, 15C from G.K. Batchelor's "An Introduction to Fluid Dynamics", in Appendix 1 (p. 594). (Every physics book is "intro to" something. I'd hate to see how hard a textbook that wasn't "just an introduction" was.
#define X(x,y) x##y
Peter Cordes ; e-mail: X(peter@cordes ,
You might want t check this out a little further - in terms of vision, for instance, initial processing of images from right eye is done by the left side of the brain, and vice versa. (The wiring is strangely crossed). Of course, different functions that use visual info are on either the right or left side, respectively. Because of the bandwidth choke between hemishpheres, info from different eyes will get to functional areas of the brain at different rates.
As you might imagine, some visual tasks are traditionally "right brain" and some are "left brain." So, in tasks where subjects were required to use only one eye, for instance the left eye, they would do better at "right brain" activities. Cover up the other eye, and they would do better (faster) in "left brain" tasks.
And again, the reason is exactly the same as in processors. Info transfer is MUCH faster within hemispheres as is is between them. Similarly, info transfer within a chip is much faster than the front side bus speed, which is the rate at which info would transfer between chips. So one fast chip is always preferable to two slower chips if I have a fixed amount of transistors to work with.
And our brains would work better, too, if it weren't for that info choke between hemispheres. That's one of the disadvantages of bilateral symmetry in humans.
-Looking for a job as a materials chemist or multivariat