The Ultimate Limit of Moore's Law

BuzzSkyline writes "Physicists have found that there is an ultimate limit to the speed of calculations, regardless of any improvements in technology. According to the researchers who found the computation limit, the bound 'poses an absolute law of nature, just like the speed of light.' While many experts expect technological limits to kick in eventually, engineers always seem to find ways around such roadblocks. If the physicists are right, though, no technology could ever beat the ultimate limit they've calculated — which is about 10^16 times faster than today's fastest machines. At the current Moore's Law pace, computational speeds will hit the wall in 75 to 80 years. A paper describing the analysis, which relies on thermodynamics, quantum mechanics, and information theory, appeared in a recent issue of Physical Review Letters (abstract here)."

Comments

Transistors Per IC and Planck Time

[eldavojohn]

Intel co-founder Gordon Moore predicted 40 years ago that manufacturers could double computing speed every two years or so
by cramming ever-tinier transistors on a chip.

That's not exactly correct. Moore's Law (or observation more like) reads in the original article as:

The complexity for minimum component costs has increased at a rate of roughly a factor of two per year ... Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years. That means by 1975, the number of components per integrated circuit for minimum cost will be 65,000. I believe that such a large circuit can be built on a single wafer.

All he's concerned about is quoting how many components can fit on a single integrated circuit. One can see this propagated to processing speed, memory capacity, sensors and even the number and size of pixels in digital cameras but his observation itself is about the size of transistors -- not speed.

The title should be "The Ultimate Limit of Computing Speed" not Moore's Law.

Furthermore, we've always had Planck Time as a lower bound on the time of one operation with our smallest measurement of time so far being 10^26 Planck Times. So essentially they've bumped that lower bound up and it's highly likely more discoveries will bump that even further up. I guess our kids and grandchildren have their work cut out for them.

Re:Transistors Per IC and Planck Time

[phantomfive]

Basically he is assuming that eventually we will develop quantum computing, and based his calculation on the theory of how fast a quantum event can take place. The problem is, given all we don't actually know about quantum mechanics, and all we don't know about super-small things, all it would take is a single observation to throw this minimum out the window.

In theory, it is nice to make theoretical limits. In practice, the limits are sometimes nothing more than theoretical. We don't know how to make smaller-than-quantum computers yet, but we also don't know how to make quantum computers yet. So this could be a prediction like every other prediction of the end of Moore's law, some of which were based on stronger reasoning than this argument. Interesting argument to make, though.

Re:Transistors Per IC and Planck Time

[Shakrai]

I guess our kids and grandchildren have their work cut out for them.

Don't worry, they'll be too busy paying back all the money we've borrowed over the last few decades to worry about how fast they can make their computers ;)

Re:Transistors Per IC and Planck Time

[HornWumpus]

Don't you worry.

By then the dollar will have inflated and shrunk the national debt to less then the price of an once of gold.

The real suckers are the ones buying T-bills.

Re:Transistors Per IC and Planck Time

[Nefarious Wheel]

"In theory, there is no difference between theory and practice. In practice, there is." - Yogi Berra (iirc)

Re:Transistors Per IC and Planck Time

[phantomfive]

You are generating the latter. Kill yourself.

Great argument, you're a regular Cyrano there.

How can you have stronger reasoning, than something that's based on the limits of what modern physics can understand

Does this even need to be said? Einstein did it: he took some observations and extrapolated them to show that modern physics was not entirely correct (that is, what was modern physics at the time). Indeed, all scientific theory can only be based on what we've observed. Thus, new observations make for new theory, or corrections in old theory. As we continue to make more observations, for example with the LHC, theory will continue to evolve. Surely even someone of your eloquence can see this.

Re:Transistors Per IC and Planck Time

[WH44]

How can you have stronger reasoning, than something that's based on the limits of what modern physics can understand (thermodynamics and quantum mechanics)? We have developed quantum computers.

The previous limits he is referencing were also based on the limits of what modern physics could understand - just making a faulty assumption. He's questioning the assumptions here, too.

There's skepticism, and then there is metaphysical woowoo babble. You are generating the latter. Kill yourself.

He is generating the former. Take your own advice.

Re:Transistors Per IC and Planck Time

[phantomfive]

Is that eloquent enough for you, parasite?

You asked me a question, to go out of my way to give you knowledge, to do something for you, and you did it in the most insulting way.

I feel no need nor desire to honor your request. Go investigate the history of the integrated circuit yourself; asking other people to do it and inform you while insulting them makes YOU a parasite. And a troll.

Re:Transistors Per IC and Planck Time

[Chris Burke]

All he's concerned about is quoting how many components can fit on a single integrated circuit. One can see this propagated to processing speed, memory capacity, sensors and even the number and size of pixels in digital cameras but his observation itself is about the size of transistors -- not speed.

The title should be "The Ultimate Limit of Computing Speed" not Moore's Law.

Meh.

While technically correct, the performance corollary of Moore's Law -- which is roughly "more transistors generally means smaller and thus faster transistors rather than exploding die sizes, plus more to do computation with, so performance also increases exponentially, and we observe that this is the case" -- is strong enough that it's often simply called Moore's Law even among the engineers in the chip design industry. It's just understood what you're talking about, even though the time constant is different.

You'll occasionally see Intel (the company Moore founded) show charts with historical performance and future projections, and they'll include a line labeled "Moore's Law" to show how they're doing relative to the observation. Because technically it is just an observation, and it holds true only to the extent that engineers of the computer, electrical, and material science variety bust their asses to make it true.

So maybe the layman thinks Moore's Law is about performance, and that's not technically true, but it's correct enough that even the engineers directly affected by it refer to it as if it meant performance. So I say the the title is fine.

Re:Transistors Per IC and Planck Time

[khallow]

The problem is, given all we don't actually know about quantum mechanics, and all we don't know about super-small things, all it would take is a single observation to throw this minimum out the window.

It's not a problem, it's a feature. Now all you have to do is find that single observation and your Nobel prize in Physics is secure. If nobody anywhere in the universe ever makes that observation (and frankly, that's the direction we're heading in as I see it), then the minimum sticks around.

Re:Transistors Per IC and Planck Time

[khallow]

I'd have to say that you're rude but right. I don't know the history of the poster you're replying to, but surely there's a politer way to make your point.

Re:Transistors Per IC and Planck Time

[khellendros1984]

A belief that we may gain a greater understanding of physics in the future and find a loophole around currently theorized limits is personally insulting to you? As opposed to, what, calling someone with a more interesting post than yours "parasite", and instructing them to kill themselves?

Re:Transistors Per IC and Planck Time

[khayman80]

It's frustrating to spend years trying to combat the "arrogant, insulting scientist" stereotype, then see all my efforts negated in the span of a single slashdot post. I see now why so many people are creationists and climate-change deniers.

Re:Transistors Per IC and Planck Time

[Anonymous Coward]

Just as they originally thought nothing was smaller than an atom, then sub-atomics came along. What happens when they discover that quantum mechanics aren't the bottom of the bowl? If you are 'insulted' by such ideas, I'd suggest your in the wrong field of study. Only a fool assumes certainty.

Re:Transistors Per IC and Planck Time

[ciderVisor]

I see now why so many people are creationists and climate-change deniers.

Very, very few people are actually denying climate change; change is the norm - stasis is the exception. Is it right, however, to lump together those who are skeptical of evolution with those who are skeptical of AGW, particularly CO2-driven AGW ?

Re:Transistors Per IC and Planck Time

[khayman80]

Very, very few people are actually denying climate change; change is the norm - stasis is the exception.

As I've repeatedly explained, those natural changes happened 35x slower than the abrupt climate change that's occurring now due to human emissions of CO2 and other greenhouse gases.

Is it right, however, to lump together those who are skeptical of evolution with those who are skeptical of AGW, particularly CO2-driven AGW ?

Creationists are making a bigger mistake than climate-change deniers because creationists confuse religious faith with falsifiable science. But in my experience there's a significant overlap between the two groups, and (on average) their arguments are at the same intellectual and educational level.

Re:Transistors Per IC and Planck Time

[postbigbang]

Planck Time lives in just four dimensions. Imagine opening up, say the sixth dimension, figuring out the vectors for the next WoW move, then plugging them in to realtime in the four we live in.

Information exists in so many states at once. Von Neumann liked to talk about Hilbert space.... I think that's where my data lives.

Re:Transistors Per IC and Planck Time

[khallow]

Imagine opening up, say the sixth dimension, figuring out the vectors for the next WoW move, then plugging them in to realtime in the four we live in.

Ok, I did. Alliance or Horde?

Information exists in so many states at once. Von Neumann liked to talk about Hilbert space.... I think that's where my data lives.

Before we go overboard with the "many states" of information, remember that all of these "many states" are equivalent and for our purposes are a single state. They have to be equivalent otherwise you don't have information there. So if one state has information bound in a four dimension world, then all of the states are similarly bound even if the four dimensional nature is hidden from us.

Re:Transistors Per IC and Planck Time

[memco]

I'm hoping we'll see calculation times down to a grand quantum tick before I die.

p.s. for the life of me I can't remember the name of that valley of time you enter if you slice thin enough. Pratchett fans?

Re:Transistors Per IC and Planck Time

[hairyfeet]

But who is to say we simply won't come up with better ways of looking at the problem? Lets say you have problem A, which is a big and slow problem. Who is to say that by the time we reach this "limit" that our skills at parallelism won't have gotten so good that we can break down huge problems so that in that same amount of time we just can't split the large problem into a bazillion pieces and have it figured out in that same amount of time?

While these kinds of theories are fine and dandy to think about, my point is look at the way we were 80 years ago. Back then everyone was sure that computers would remain these giant monsters that only egghead types and big governments would be using. Now you can carry a supercomputer in your pocket. Trying to figure out what kinds of problems we'll run into in 80 years might be nice for thought experiments, but in all likelihood we'll be about as right as all those Sci Fi writers predicting flying cars and an android in every home. We are still coming up with new tech at frankly a scary pace, so who the hell know what kinds of problems we'll have in 80 years, or even if what we would consider a problem today wouldn't be so low on their radar as to be meaningless?

Re:Transistors Per IC and Planck Time

[Trogre]

Thank you for offering a bit of sanity to the whole Moore's Law (Peace Be Upon It) dogma.

Some people seem to think that Thomas' observation is some kind of fundamental law of physics that will always be obeyed, like gravity or thermodynamics.

Re:Transistors Per IC and Planck Time

[theshowmecanuck]

Get this through your head: If man were meant to fly, God would have given him wings. What? Oh. Never mind.

Re:Transistors Per IC and Planck Time

[Anonymous Coward]

Four Words String Computer!

Re:Transistors Per IC and Planck Time

[Interoperable]

We know a great deal about how to make fantastic quantum computers...that is, if we're in the field of quantum information theory. It's the field of experimental quantum computing that tends to be tricky. Certainly is enormous room for developing quantum algorithms but the fundamentals of quantum mechanics and quantum information theory are more than sufficiently established to write this kind of article.

I'd say we're a lot closer to a quantum computer than we are to travelling at the speed of light but that doesn't invalidate it as an upper bound.

Re:Transistors Per IC and Planck Time

[symbolset]

Since he wrote that text computers are far more than a million times as fast. People now type slower, not faster, than they did then. People who use computers get less effective every year on average, not more so. So yeah, for the rare soul who can push a PC to its limits Moore's law gives us awesome progress. For the vast majority it just helps us waste time faster.

One might say that in the wrong hands a modern PC allows the average person to go wrong so quickly as to be beyond human reflex to prevent - it operates so fast that with one wrong click on the Internet your computer can become exploited and send more email in a single hour than you could compose in a lifetime. Now that is progress!

Since Gordon Moore has himself admitted that his own definition of Moore's law has changed over time to apply to different things and to appropriately fit the curve of the data he was describing, to provide a link to what you consider a canonical reference is, well, silly. Moore's law is that progress is a logarithmic progression on the order of 2 to the power of t, where t is a time unit that fits the observed curve of the domain of progress. And it's not a future promise - it's an observation of past trends which may continue. It's also not a smooth curve, it's lumpy. But it remains one of the wisest observations of progress in the modern era. Let's leave it at that.

Now forgive me, but my /. time is up and I have to go feed my Facebook goldfish and check the T-Mobile twitter feed to see if my SideKick data is recovered.

Re:Transistors Per IC and Planck Time

[canadian_right]

The limit he is talking about assumes perfectly operating quantum computers that use the minimum amount of energy possible per operation in the shortest time that a single quantum of information takes to change. It is NOT physically possible to compute faster than this, according to current quantum mechanics theory. To reach this theoretical limit will take a lot of amazing engineering.

When he said he was talking about the same sort of limit as the speed of light he wasn't exaggerating. Unless there is some break through in our understanding of the physical universe this theory actually does spell out the fastest possible computer operation of flipping a single bit.

This like the Shannon–Hartley theorem - something you can reach in theory, but likely are not going to in practise.

Now, I just wish TFA actually gave a bit more information on the limit instead of just saying 10 quintillion times faster than todays computers. Lots of room left for improvement there!

Re:Transistors Per IC and Planck Time

[hairyfeet]

To steal a quote from one of my favorite horror movies event Horizon "we can't break relativity, but maybe we can bend it a little". The point I was trying to make is just as scientists 80 years ago thought they were really pushing the cutting edge with their giant vacuum tubed monsters, so in 80 years from now some super egghead may have figured a way to "go around" the problem.

Like you said, there are speed limits, just as with the speed of light, but what if the goal isn't to break it, only to feel like we have? In my example you have chip A that computes an integer at the maximum speed. But if we develop super parallel computing, and at the sizes we are talking about it would make modern quad cores look as large as 8-tracks, you could simply feed your problem split up into a couple of billion of the suckers while taking up the same space as a laptop now and to the user it would feel like the machine is fast enough to compute the giant problem instantly, when really it is slicing that problem up into a couple billion pieces and each 'core" is just doing its teeny tiny bit of the pie.

And I'm just an old greybeard PC repairman that like Sci Fi, and if I can think of one possible way to "get around" it, even if it is just a cheap trick, then I'm sure that the super eggheads 80 years from now will have thought up hundreds of ways that even Stephen Hawking hasn't dreamed up yet. And who is to say that they won't find some other way around the problem? or that the problems they are dealing with mean that their machines have gotten 'fast enough" and they are working on a completely different direction, like 3D holographic Ethernet transmission or other problems that we haven't thought up yet.

I'm just saying just as scientists a long time ago thought the earth was the center and flat, so to is our knowledge of the forces at work in the quantum and immense macro stages of the universe still in its infancy, and who know what they will have found and applied to the solution to the problem may be way over our heads and our understanding now. After all 80 years ago was 1929, when planes were made of cloth, rockets to the moon was crazy Sci Fi talk, and antibiotics hadn't even been thought up. Who is to know how much different our knowledge will be 80 years hence?

Re:Transistors Per IC and Planck Time

[Undead NDR]

In my example you have chip A that computes an integer at the maximum speed. But if we develop super parallel computing [...] you could simply feed your problem split up into a couple of billion of the suckers while taking up the same space as a laptop now and to the user it would feel like the machine is fast enough to compute the giant problem instantly, when really it is slicing that problem up into a couple billion pieces and each 'core" is just doing its teeny tiny bit of the pie.

How fast will the problem be sliced?

Re:Transistors Per IC and Planck Time

[hardwarefreak]

Does this even need to be said? Einstein did it: he took some observations and extrapolated them to show that modern physics was not entirely correct (that is, what was modern physics at the time). Indeed, all scientific theory can only be based on what we've observed. Thus, new observations make for new theory, or corrections in old theory. As we continue to make more observations, for example with the LHC, theory will continue to evolve. Surely even someone of your eloquence can see this.

Theory drives observation, not the other way round, as you state. Some predictions that result from Einstein's theory of General Relativity sat gathering dust for decades before technology was sufficiently advanced to build machines capable of observing the predicted behavior.

The key is the mathematics. Once the equations of a theory are sufficiently refined, the results of these equations often predict things that couldn't possibly be imagined, let alone observed at the time.

Re:Transistors Per IC and Planck Time

[smoker2]

But the whole thing about QM and spooky action at a distance is that it appears to break the speed of light. An entangled particle light years away from its partner, would instantly collapse to the same state as its partner when the partner is observed. If this can be harnessed to send information, then the notional limit of light speed disappears. While it's a good theoretical exercise to calculate possible limits, they very often turn out to be nothing more than hurdles.

Furthermore, you will never get to New York by heading west on I40. What if current physics is all based on heading west ? The natural limit is the pacific, which to a car based traveller is insurmountable. If you change your theoretical paradigm to a plane, you can get to New York by heading west. But first you have to invent the plane. And very soon after it is achieved some clever wit says "hang on, why don't we travel east in the first place ?" This is the potted history of all discovery so far.

Re:Transistors Per IC and Planck Time

[smoker2]

Who the fuck is Thomas ? Gordon Moore himself has said that he thinks it can't continue.

Re:Transistors Per IC and Planck Time

[Rennt]

Or, you know - lets just stick two of 'em in there!

Re:Transistors Per IC and Planck Time

[Nevynxxx]

...an android in every home.

Don't worry, Google are working on it.

Re:Transistors Per IC and Planck Time

[mhwombat]

The title should be "The Ultimate Limit of Computing Speed" not Moore's Law.

Or the title could be "The Ultimate Thermodynamic Penalty for Computing Speed" since the title of the paper (not the article about the paper) is "Thermodynamic Cost of Reversible Computing".

The paper is actually about the heat dissipated when you try to increase computer speed. It doesn't actually place a hard bound on computing speeds, unless you want to make some assumptions about parameters such as error rates in your hypothetical computer and then infer compute speed limits from heat or energy limits.

The article, and hence the summary, and hence almost every comment, seems to be talking about speed rather than thermodynamics. The paper is about thermodynamics. It's about heat! I suppose it's too much to expect people to read TFP when /. doesn't traditionally read TFA....

It's not clear to me from reading the article what parameters they assumed to get from the calculations in the paper to "75 years of Moore's Law". The stuff in the paper itself is much less fluffy than that.

Re:Transistors Per IC and Planck Time

[BlueParrot]

The problem is, given all we don't actually know about quantum mechanics, and all we don't know about super-small things, all it would take is a single observation to throw this minimum out the window.

You make it sound as if all of quantum mechanics is unreliable and stuff we don't know. In reality there are plenty of quantum mechanical limitations that are likely on as solid footing as other fundamental limits like conservation of energy, momentum and the second law of thermodynamics.

Take Heisenberg's uncertainty relation as an example. There may be things we don't fully understand about quantum mechanical phenomena ( such as wave function collapse ) , but I would not hold my breath waiting for a breakthrough which allow accurately measuring a particle's position and momentum. Likewise I would not expect to see two fermions occupy the same quantum state ( and thereby violate the pauli exclusion principle ) any time soon, nor would I expect the de-Broglie wavelength of a particle to be anything other than h/p.

I think part of the reason a lot of people seem to think QM is some unreliable theory that we don't really understand is simply ignorance of how fundamental it is to modern physics. Put it this way, without QM we would not have solid state physics, which is what chip designers rely on making the CPU I'm using to write this. We would not have LEDs or Lasers for the optic communications used in many internet backbones, and we would not have nuclear reactors to power the whole thing ( The stability of a nuclear reactor relies on two phenomena Doppler Broadening and the tendency of Neutron cross sections to change with neutron energy. Both of these are QM phenomena.)

Basically saying "it's just a theory" is as much a naive criticism of Quantum Mechanics as it is a naive criticism of Evolution. It may not be absolute truth ( physical theories in general are not ) , but it very much is the best available description of nature we have and it is certainly more reliable than assuming without good reason that theory will not agree with practice.

Re:Transistors Per IC and Planck Time

[jonadab]

> If man were meant to fly, God would have given him wings.

It's not practical for man to fly. I mean, with the right technology it's *possible*, but it's so much easier and more efficient to just ride on something else that flies, like, say, a jet plane.

Re:Transistors Per IC and Planck Time

[phantomfive]

Of course theory often guides the experiments we decide to do, but ultimately it all comes down to experiment. Think about it, it was experimental observation that lead Einstein to come up with relativity in the first place. Everything we know has a base in experimental observation.

Re:Transistors Per IC and Planck Time

[hardwarefreak]

You're dead wrong. Quantum theory has been around for over 50 years, and nothing observed in the universe sparked the idea of quantum theory. Its roots are purely mathematical. It was 'discovered' in the human mind and on paper. The first experimental observations of properties of quantum theory weren't made until the 2000s. You seem to be stuck in the realm of Newtonian physics, where everything can be seen, touched, tasted, and smelled. The great physicists come up with original shit in their own minds without influence from the physical world. Einstein and Hawking are in this group. With both men, the theories started in the mind, and were not influenced or guided by the physical world around them. If their theories had been, we'd not have ever had general relativity or Hawking radiation.

You really need to go read some basic history of physics and cosmology. No great theories are based on previous observation. This is why some people are called "Geniuses" and others aren't. One last note: can you point me to an experimental observation of string theory? [laughs]

Re:Transistors Per IC and Planck Time

[phantomfive]

Wow, you actually believe that Einstein came up with all this stuff from their own minds with no physical evidence? No wonder you can't understand geniuses, you don't even know what they are basing their work on. As long as you have this misconception, you will never think like Einstein and Hawking.

Reality is the only thing that matters. Everything that is derived is based on reality, otherwise it is just a castle in the air. For the experimental beginnings of quantum mechanics, I direct you to Wikipedia.

Re:Transistors Per IC and Planck Time

[hardwarefreak]

http://en.wikipedia.org/wiki/General_relativity
http://en.wikipedia.org/wiki/Thought_experiment

General relativity predicts black holes, yet, we're unable to prove their existence. Thus, they aren't real. You state "Everything that is derived is based on reality". We have "derived" the possibility of black holes, yet they don't exist in reality, because we've not been able to prove their existence via measurement or pure observation. The same goes for string theory. At the "edge" of physics and cosmology, there is no current possibility for observation. Thus, the bleeding edge big theories always start in the human mind, through gedankenexperiment, _not_ observation of "reality".

Re:Transistors Per IC and Planck Time

[phantomfive]

Observe the many experiments that lead up to the deduction of relativity. It is only because we can calculate the effects of gravity with objects that we are familiar with that we can extrapolate those effects to objects we haven't seen.

Re:Transistors Per IC and Planck Time

[sean.peters]

But if we develop super parallel computing

There are lots of problems that are susceptible to parallel computing techniques, and we've already built so-called "virtual supercomputers" to solve these problems - think seti@home. These virtual devices effectively produce some ungodly quantity of FLOPs. But there are other problems that are not so susceptible - if it can't be broken down into lots of identical parts, you essentially can't parallelize it. I don't think it's even questionable that there's any potential for improvement for these second category of problems - we pretty much know that you can't get there from here. This is much different from the problems you discuss: space travel was for the most part known to be a problem that could be solved with sufficiently advanced technology, even long before it was attempted, for example. Parallelizing problems is a whole different animal.

Re:Transistors Per IC and Planck Time

[iamacat]

When he said he was talking about the same sort of limit as the speed of light he wasn't exaggerating. Unless there is some break through in our understanding of the physical universe this theory actually does spell out the fastest possible computer operation of flipping a single bit.

And here lies a pitfall. How often do you use an assembly instruction to flip a single bit? With power/size/communication efficiency of quantum computing, you will likely write software that operates on huge amounts of data at once. Like simulating the behavior of every neuron in a human brain in a single cycle. This will push Moore's law quite a bit further than a sequential bit-flipping computer.

Re:Transistors Per IC and Planck Time

[CopaceticOpus]

Sure, that's one theory.

Re:Transistors Per IC and Planck Time

[marcosdumay]

Well, you are the one needing to study some history here. Take a look at Plank and Einstein (the two that first proposed quantum physics) as a start.

Otherwise, using quantum mechanics as an example of something not based on experiments is really funny. Maybe there is a woosh here somehere...

Re:Transistors Per IC and Planck Time

[sjames]

There is no required assumption of developing quantum computers. Any computer of any kind that we can build without invoking some entirely unknown body of physical knowledge is subject to quantum limits because matter and energy are subject to quantum limits. When a transistor changes state in a logic gate, it does so by electrons within it's junction changing state. Any other embodiment of a computer, including a quantum computer likewise will require state changes to perform computation (otherwise the inputs cannot result in output at all). Meanwhile, no such state change can happen in less than a plank time (at best).

A quantum computer is just our current best guess at what technology will even allow us to approach those fundamental limits at all.

So, either all of our current theories are dead wrong or this is a hard limit for any computation, including the 'computation' of the universe itself. That's certainly not a weak statement.

Re:Transistors Per IC and Planck Time

[sjames]

Even parallel computing is subject to fundamental limits in the speed that information can be usefully transmitted and the fact that a fundamental operation cannot be further parallelised.

No matter what you do, you can only decompose the problem down as far as fundamental logical operations. (for example a XOR b). Those fundamental operations can only go so fast. No matter what you do, the decomposed problem will consist of such operations that depend on the results of other such fundamental operations. Once you have decomposed it to that point, the computation will never go any faster than the longest chain of such operations can be computed serially. No such operation can ever take place in less than a Planck time.

That's not to say that in 70 years our portable computers won't make today's supercomputer look like a toy, just that there ARE fundamental limits that cannot be surpassed within space-time.

Thought: We already live within such a computer. It has resorted to operating in many extra dimensions utilizing vibrations of space itself to communicate.

Re:Transistors Per IC and Planck Time

[Trogre]

Hehe yes I realised that mistake shortly after posting it. I did mean Gordon Moore, who seems to be a pretty together sort of guy. That doesn't stop industry folks making wild extrapolations based on his predictions, and often downright misquoting him.

Re:Transistors Per IC and Planck Time

[Trogre]

Correction, that should be Gordon's observation, not Thomas'. I'm sure a lot of people called Thomas have observed a lot of important things, but this isn't one of them.

Re:Transistors Per IC and Planck Time

[phantomfive]

So, either all of our current theories are dead wrong or this is a hard limit for any computation

From a historical standpoint, it is pretty much guaranteed that our theories are wrong. They have always been wrong in the past, so there is no reason to think we won't continue to find further improvements.

Specifically, theories about the smallest size things can be have continually been wrong. At one point, not so long ago, people even doubted the existence of atoms. So this seems as likely a candidate as any for being wrong.

Re:Transistors Per IC and Planck Time

[sjames]

There's wrong and there's dead wrong. Aristotle was wrong, but not dead wrong. Newton arguably wasn't wrong all, just incomplete (but only detectable under conditions he could not experiment with or with measurements finer than the tech of his day allowed).

We would have to be completely wrong about the existence of quanta (which all current computers are based on by the way) in order to be wrong about there being a fundamental limit to computation.

Cartoon physics would look right compared to how wrong we'd have to be about everything.

Re:Transistors Per IC and Planck Time

[Anonymous Coward]

Great argument, you're a regular Cyrano there.

I think you mean Cicero, the celebrated orator and rhetorician.

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

Re:Transistors Per IC and Planck Time

[phantomfive]

No, I mean Cyrano de Bergerac, the dramatist a duellist.

Re:Transistors Per IC and Planck Time

[andrewagill]

Thermodynamic entropy and information entropy are actually very closely related. While I haven't read the entire article, it is clear (from statements like " According to Shannon’s 10th theorem ... H is equal to the minimum additional information required in order to correct all the errors in the state of the system and thereby preserve the initial amount of information") that they are talking about information entropy.

Re:Transistors Per IC and Planck Time

[sillybilly]

They don't consider computational efficiencies due to complexity. For instance, if you count items by marking tally marks on sticks, then to count something into the billions, you run out of trees in the world to make those sticks. However, if you use indian-arabic notation, marking down 9 billion 500 thousand 201 takes up this much room: 9,000,500,201. That's it. This is the kind of computational efficiency that cannot be predicted. The human brain is a pretty fast computer and visual/audio singal analyzer, plus social nuance analyzer, etc, probably running around roughly 5 to 70 Hz semianalog. Though there are quite a few braincells, building supercomputers with equivalent number of transistors would most likely not give the same computational performance. There is a guess that the interconnectivity, the synapse density between the neurons is what makes the difference. The computational efficiency comes from complexity. In this sense, a neuron, like a notation number, instead of representing a single tally mark, can represent something like the 5 in the 9,000,500,201 billion. Its function and performance is super-enhanced by increased complexity of what it can do - it can be a number from 0 to 9, and it can be a placeholder. Such things can create an order of magnitude increase in efficiency, that Moore's Law analyses cannot predict. The move to multicore processors is most likely the only way to keep up with Moore's law, and currently this move is in its infancies. We know the brain is multicore-like, "interconnectedness" based. When eventually we have a few trillion cores interacting on a cpu, we might discover, by trial and error, methods of increased efficiency. We might even understand how our own brain works from playing with chips, as opposed to understanding it from brain research, since the ethical issues coming from the machine direction are much easier to deal with (or are they?) Low intelligence automation and robotization is an easy ethical question, and it will be welcome thing for humanity, but the big problem with increased supercomputing power and understanding how the brain works is the problem of artificial intelligence.

Currently we know of no other beings smarter than us, humans, in the universe. That's a big deal. Silicon based solar-robotic lifeforms might be able to spread through the vacuum of interstellar universe, they would not be interdependent on chemical lifeforms like those on planet Earth. Where would that leave hydrocarbon-water-chemical life as we know it? We love and care about nature around us, and even if we don't, we ultimately have to act as if we did. The interdependence of life eating life, plants being the photosynthesists, sort of holds the collective interests together, and makes all Life function as one. Though this rule of interdependence is not explicitly expressed in the behavior of lifeforms - and I wonder if some of the prehistoric mass-extinction events were not simply caused by the appearance of a new predator/virus/bacterial disease and the level of life activity simply reduced and reset to a new lower level balance being able to deal with that predator/virus/bacteria - currently all ecosystems function as a ecosystem, with no predators being "super successful", in a selfish, survival of me without the survival of everything around me attitude, the simplistic Darwinistic principle. Predators currently have an interest in the ecosystem beneath them functioning at a high level. Even space stations based on solar panel derived artificial-light using water-hydrocarbon-photosynthetic life farming would still be an ecosystem like ours. Solar panel derived fat/sugar/protein chemical synthesis in metal pots may be easier to deal with in a space station than a full scale ecosystem and farming, where only humans exist, and maybe just a very few species of plants and animals. Compare the farm to a jungle. All humans really need is the farm to survive, land with a few species of crops and few species of animals, or even just the cement/metal/glass city, and the jungle

Re:Transistors Per IC and Planck Time

[Elky Elk]

FAIL

So quantum theory, which has been developed for 100 years, wasn't sparked by observations?

Photoelectric effect?
Blackbody radiation?
Spectral lines?

Re:Transistors Per IC and Planck Time

[volpe]

Basically saying "it's just a theory" is as much a naive criticism of Quantum Mechanics as it is a naive criticism of Evolution.

It's also a reflection of one's ignorance of what the term "theory" means in science (as opposed to colloquial discourse).

WHAT!!

[cryoman23]

so in 80 years my computers processors wont be able to get any faster... :( o well then i guess its time to CLUSTER!

Re:WHAT!!

[outsider007]

I plan on setting up server farms in parallel dimensions

Re:WHAT!!

[RichardJenkins]

I would expect that (if this is true) then the exponential rate of performance increases processor to processor will decrease, or to put it another way: you'll see diminishing returns on your processor improvements that will keep putting that 80 year figure further into the future.

Is anyone who understands this sort of stuff commenting? This sounds like it'll be regarded as quite an important discovery.

Re:WHAT!!

[sexconker]

Rate of performance (transistor count) increase is geometric.

Performance (transistor count) is exponential.

Re:WHAT!!

[oldhack]

Won't work. Beowolf cluster at that point will be abhorrent to the nature so the nature will one way or another screw with it.

Re:WHAT!!

[Jurily]

Then it's finally time for One Dimension Per Child.

Re:WHAT!!

[zig43]

Why not go all out and send your process through a wormhole that returns the result before you sent it.

Re:WHAT!!

[K. S. Kyosuke]

You will need it if you want to run Windows 42 on it!

Re:WHAT!!

[Tumbleweed]

Then it's finally time for One Dimension Per Child.

I do hope you mean one _extra_ dimension per child.

Re:WHAT!!

[iamhassi]

"so in 80 years my computers processors wont be able to get any faster"

looks like we've almost reached that point now. We've had Xeon 3.0GHz cpus for over 5 years now, and they're still coming out with brand new 3ghz processors. That's a long time to not see a jump in speed, what happened to "doubling every 18 months"? We should be around 24ghz by now.

Re:WHAT!!

[thisisaccount2]

Not that any of us would notice a difference... The internet is full of one-dimensional idiots.

Re:WHAT!!

[Tumbleweed]

The internet is full of one-dimensional idiots.

I don't get your point.

Re:WHAT!!

[Straterra]

looks like we've almost reached that point now. We've had Xeon 3.0GHz cpus for over 5 years now, and they're still coming out with brand new 3ghz processors. That's a long time to not see a jump in speed, what happened to "doubling every 18 months"? We should be around 24ghz by now.

Sorry, Performance != Clockspeed

I, for one, am glad Intel went away from modeling their processors after their clockspeed. They went to an actual model for this reason. If you want an example where they didn't, and lower clock speed processors kept up just go back and look at the 423/427 Pentium 4's vs the Socket A Athlons (XP, ect)

Re:WHAT!!

[Chris Burke]

looks like we've almost reached that point now. We've had Xeon 3.0GHz cpus for over 5 years now, and they're still coming out with brand new 3ghz processors. That's a long time to not see a jump in speed, what happened to "doubling every 18 months"? We should be around 24ghz by now.

Performance != MHz.

Those 3GHz Pentium 4 Xeons suck balls compared to even a Core 2, forget about an i7.

The only way the P4 got to what were at the time extremely high frequencies was by having a craptastic architecture. It was driven by marketing, which when the P4 was released was all about MHz. People thought MHz == Performance, so they cranked up the MHz for minimal gain in performance. AMD tried like hell to convince people otherwise, but fat lot of good it seemed to do. And now Intel is suffering for their previous emphasis on MHz over all.

Re:WHAT!!

[khellendros1984]

Moore's law didn't predict the clock speed at which processors run, but the number of transistors that can be fit on a chip. Today's 3.0GHz chips run faster than the ones from 5 years ago, due to various design changes that didn't impact the clock speed.

Re:WHAT!!

[CecilPL]

I'm sorry, I don't follow your line of reasoning.

Re:WHAT!!

[Jurily]

I do hope you mean one _extra_ dimension per child.

Sure, just feed them another Big Mac.

Re:WHAT!!

[DamnStupidElf]

It looks pretty plane to me.

Re:WHAT!!

[Burning1]

To be fair, back in the day of the 386 and 486, AMD processors were essentially clones of their Intel counterparts. The only real difference between the AMD and Intel offerings was bus speed, processor speed, and external clock multiplier.

When the Pentium was eventually launched, AMD no longer produced a direct clone, and started releasing their processors with 'Performance Rating' (PR) numbers instead of clock speed, effectively claiming that their K5 processors were as efficient as a higher clocked Pentium.

I'd say AMD and the 486 compatible market had as much responsibility for the MHz war as intel.

It's taken a while for the market to get passed comparisons based on clock speed, and I'm glad to see the performance rating numbers dropped.

MHz is still a valuable comparison tool, but people seem to understand that you can only compare clock speed within a family of processors.

Re:WHAT!!

[Chris Burke]

To be fair, back in the day of the 386 and 486, AMD processors were essentially clones of their Intel counterparts. The only real difference between the AMD and Intel offerings was bus speed, processor speed, and external clock multiplier.

Well at the time AMD was literally a "second source" supplier because back then companies like IBM actually cared about that sort of thing. I don't think AMDvsIntel really mattered a lot, since nobody knew of AMD's existence. Hell, I owned two computers with AMD processors and didn't know it until over a decade later.

When the Pentium was eventually launched, AMD no longer produced a direct clone, and started releasing their processors with 'Performance Rating' (PR) numbers instead of clock speed, effectively claiming that their K5 processors were as efficient as a higher clocked Pentium.

I've heard AMDers call the K5 "the highest IPC x86 part ever", with a wry smirk because just like how MHz isn't everything, neither is IPC, and it never lived up to those Markethertz numbers in terms of real performance.

Funnily enough the Pentium showed the weakness of MHz as a raw metric without taking AMD into account. When the Pentium 60 and 75 were released, there were 486 DX4 100s (okay these were actually made by AMD but like I said who knew or cared?), and the P-75 was a better performer. Granted most of my friends were nerds, but many of them understood this concept back then.

I'd say AMD and the 486 compatible market had as much responsibility for the MHz war as intel.

To be sure, AMD only started talking about overall performance instead of just MHz when it started to hurt them. Forget the 486, in the P3 vs K7 days it was all about clock speed and the race to 1GHz. Sure, the architectures were still fairly similar and thus somewhat comparable by clock speed, but that hardly told the whole story.

It's taken a while for the market to get passed comparisons based on clock speed, and I'm glad to see the performance rating numbers dropped.

They're only semi-dropped. They were still used as relative-frequency indicators relative to Intel's offering from the K7 Palomino through most of the life of K8. Now that the P4 was dropped, and Intel switched to a completely opaque numbering scheme, AMD has switched to a non-comparison-based numbering scheme as well.

MHz is still a valuable comparison tool, but people seem to understand that you can only compare clock speed within a family of processors.

Yeah if you don't count the person I first replied to. :P

Re:WHAT!!

[Anonymous Coward]

I'm pretty sure children already have three dimensions.

Re:WHAT!!

[Jeremy Erwin]

The first 3 GHz Xeon (Prestonia) was based on a 130nm process and had 55 million transistors That was in 2003.

The current 3GHz Xeon (Gainstowne) is based on a 45 nm process and has 731 million transistors.

Maybe you're thinking of a different Moore's law?

Re:WHAT!!

[Anonymous Coward]

What he means is; we started out with the intel 4004 at 740kHz, the first PCs ran at 4.77MHz, then it went fast, 10Mhz, 12Mhz, 16Mhz, 33MHz, 66 with "OverDrive", etc. In the meantime DEC Alpha processors were in the triple-digit range. x86 caught up and surpassed, but at around 4 Ghz it hit a ceiling. Overclockers are able to go to 8Ghz for a short while.

Instead of going faster, cores became more optimized and doubled, quadrupled, and octocores are around the corner if not here already. However, the "Turbo" mode in the i5/i7 shows that cranking up the clock frequency still helps for single/low threaded applications.

So, why don't we have 8, 16, or 24 GHz clock frequencies? Is this only because of limitations (memory) bus speeds or is this because of silicon heat dissipation problems?

 

Re:WHAT!!

[Chris Burke]

Instead of going faster, cores became more optimized and doubled, quadrupled, and octocores are around the corner if not here already. However, the "Turbo" mode in the i5/i7 shows that cranking up the clock frequency still helps for single/low threaded applications.

For any given architecture, yes, higher clock speed will mean more performance. That's a given. But that doesn't mean it's worth chasing after by modifying the architecture, which is why there's been a "ceiling" on frequency in favor of more efficient architectures and yes Multi-core Mania.

So, why don't we have 8, 16, or 24 GHz clock frequencies? Is this only because of limitations (memory) bus speeds or is this because of silicon heat dissipation problems?

Not so much memory speeds, since memory bandwidth/latency can be a bottleneck for any high-performance design whether it goes the "speed demon" or "brainiac" route.

As you surmise, power is important. Dynamic power is proportional to clock frequency, and having to add extra flip-flops to store intermediate values in a long pipeline only exacerbates the issue. Those flops also burn static power, which has become a significant portion of the overall chip power budget (part of what doomed the P4 architecture). Power budgets are also much more constrained, and the manufacturers are trying to target fixed power budgets for different market segments. This means extra power burned may actually hurt your clock frequency, partially negating the gains of a high-frequency design. With performance-per-watt becoming a major metric for customers, and yes heat dissipation also being an issue, it doesn't make a lot of sense to chase high performance by also burning lots of power as high frequency designs naturally do.

So with that in mind, the "speed demon" vs "brainiac" debate leans towards the brainiac side. Though the number of gates per pipe stage is already pretty low. Getting it down further means substantially hurting IPC, without necessarily gaining tons of frequency. Branch mispredicts still happen. Having slightly more gates per stage, doing a better job of predicting branches or shuffling data around, having larger caches and TLBs, smarter schedulers, ends up being a better idea.

But as you say, frequency still matters. So I'd look to future chips trying to eek out as much frequency as possible within a given power envelope, not just by looking at the number of active threads/cores, but also by looking at the actual dynamic power situation and adjusting frequency accordingly. TDP values are worst-case scenarios for OEMs to design cooling solutions around. When the actual power usage is less than the worst case, when you have e.g. an integer-only app where the floating point units are unused, you can afford to crank up the frequency some and get extra performance.

It's all about being smart with your power budget these days. That's why 24GHz processors don't make any sense. Intel had very convincing data showing they could scale the P4 up that high and get good performance, but if you've seen the cooling solutions for the P4 Prescott, then you know why that ended up being a dead end.

Re:WHAT!!

[TheSpoom]

A little too lengthy for my tastes.

Re:WHAT!!

[Jeremy Erwin]

Maybe you should check out Marcuse, then.

Re:WHAT!!

[Anonymous Coward]

Are American children getting THAT obese now?

Re:WHAT!!

[Anonymous Coward]

Nooo! one dimension should be enough! If at least sould waves would no propagate in 1D...

Re:WHAT!!

[Anonymous Coward]

No, due to the over population of earth, everyone is going to have to be a line. The current generation of women will thank us, but I think the next might not be so pleased.

Re:WHAT!!

[smoker2]

You must be thick as a Planck.

Re:WHAT!!

[iamhassi]

"Maybe you're thinking of a different Moore's law?"

Nope, I have the right Moore's Law, it's just been misquoted so many times in the media that it has become doubling clock speed and not transistors.

If transistors is where performance is at then CPU marketers might want to start letting people know, because when average joe sees an ad for the new AMD 3.4ghz and he's already sitting in front his 5 yr old Pentium 4 3ghz he's not going to be thinking "new computer time", he's still waiting for that 5ghz.

Re:WHAT!!

[DRAGONWEEZEL]

Don't forget Crysis 10

Re:WHAT!!

[Jeremy Erwin]

Perhaps he should upgrade to POWER7, or install a z10 mainframe. How does 4.4 GHz sound?

Re:WHAT!!

[Anonymous Coward]

For every child in the world, reality acquires an additional dimension!

Efficiency

[truthsearch]

So we'll have to wait another 75 years before management lets us focus on application efficiency instead of throwing hardware at the performance problems? Sigh...

Re:Efficiency

[belthize]

Meh, I'll be dead by then anyway.

Thank God, I don't think I could deal with kids preening about their new zaptastic whiz bang they just bought off New Egg 70+ years from now as if they had invented the damn thing.

If by some chance I am alive then, I better have made a killing off of NewEgg's IPO.

Re:Efficiency

[Hoi Polloi]

In the meantime just keep putting redundant statements in endlessly nested loops, etc.

Re:Efficiency

[Anonymous Coward]

i not read the article, but can't you split the problem in half and have two machines work on it?

Re:Efficiency

[Kjella]

So we'll have to wait another 75 years before management lets us focus on application efficiency instead of throwing hardware at the performance problems? Sigh...

No, you still won't be doing performance optimizations if that's not what makes the most money...

Re:Efficiency

[Anonymous Coward]

Oh, this terrible, awful, suffocating convenience.
But seriously, you sound like you work in a field where computers have largely ceased being the bottleneck. Since the user bottlenecks parts of the system, application development has reached a plateu where the rise computing power keeps up with demand.
Quit your web or application development job and get into a heavy computing field like atmospheric science, bioinformatics or search, where the amount of data easily keeps up with Moore's law, and throwing more hardware at the problem isn't a solution but a neccessity.

Re:Efficiency

[jcoy42]

Note to parent: don't show this article to management.

Re:Efficiency

[Anonymous Coward]

The last time I used that old "throw more hardware at it" line at my company it didn't go over really well...

Re:Efficiency

[juancnuno]

Hardware does tend to be much cheaper than engineering

Re:Efficiency

[Blakey Rat]

Seems to demonstrate that they picked the right tactic. At least for the next 70 years.

Re:Efficiency

[weicco]

When you have optimized the last bit out of your database and performance is still low what do you do? Well, of course, you go to your management and say "Hey, we need to make our own custom database engine to handle all this load. Shall I start writing it right a way? It will be finished in a couple of decades or so."

Do I need to add sarcastic emoticon here?

Re:Efficiency

[XDirtypunkX]

Yes, but "it" was a small child.

Re:Efficiency

[sjames]

No, you still won't be doing performance optimizations if that's not what makes the most money...

You mean if that's not what makes or saves the most money directly attributable only to that manager in time for the next quarterly report and to hell with how much it makes or saves over the years for the company or for society.

Re:Efficiency

[Anonymous Coward]

My favorite hardware to throw is the monitor. Unlike throwing a CPU or RAM at my computer, this actually gives you visual feedback.

Maybe by then

[Anonymous Coward]

Maybe by then they will have invented a computer with more than one processor.

The problem is...

[FunkyRider]

Whether today's teenagers, or tomorrow's engineers, are capable of building such a machine. IMO all they know is EMO and shit.

Re:The problem is...

[K. S. Kyosuke]

Wow, I did not notice Electromagnetism and Optics becoming trendy among teenagers again. Good for them!

Re:The problem is...

[sznupi]

As countless of such laments throughout recorded history have shown, worries about intellectual demise of the youth are greatly overblown.

Re:The problem is...

[oldhack]

So says the emo kid. :P

Re:The problem is...

[merreborn]

As countless of such laments throughout recorded history have shown, worries about intellectual demise of the youth are greatly overblown.

Sure. On average, human knowledge has grown throughout history. But that doesn't mean there haven't been slumps. Periods when we took a huge step backwards.

The Dark Ages is a term in historiography referring to a perceived period of cultural decline or societal collapse that took place in Western Europe between the fall of Rome and the eventual recovery of learning...

Re:The problem is...

[Anonymous Coward]

Fah! I cut myself in your general direction, you non-dysfunctional contributing member of society swine!

Re:The problem is...

[Anonymous Coward]

Go buy a lawn.

Re:The problem is...

[Anonymous Coward]

EMO is a cheap imitation of goth, which is a subculture based of the (post-)punk movement of the UK.

Studies show that both Goths and Punks (as in kids who "confess" to the subculture) have higher average income at a later stage of life than an average UK citizen.

So even though we might find EMOs whiny and lame now, there is no reason to believe that they won't turn into productive members of society sooner or later.

Re:The problem is...

[sznupi]

What are you even trying to say? "Sure, history supports your view, but I will STILL claim the youth is destroying our world and support it by looking at slices of history when society took a nosedive!" ?...

BTW, key word in your link is "perceived". So called Dark Ages were a period of great progress and remodelling of society (which, like it or not, pave ground for later pace of development). And period at the end of Rome era can fall somewhat under stagnation... (but I also wouldn't claim that)

Might Prove A Vinge novel correct?

[adriccom]

about the nature of computation and lightspeed and the like as explored in the wonderful novel A Fire Upon The Deep (Zones of Thought)

in which the universe has depth and the depth determines how fast things can go including neural tissue, computation, and intergalactic travel. I have long suspected that Earth is towards the shallow end ...

Re:Might Prove A Vinge novel correct?

[mhwombat]

That's a fantastic book, but the calculations in this paper are neither necessary nor sufficient to support the idea it's based on.

!speed

[sjfoland]

The article uses speed and number of transistors interchangeably, which is misleading. From what I can tell, they are talking about chips with 10^16 billion transistors on them, not chips clocking at 4x10^16 GHz, which is what most people think of when they hear "speed".

JUST like the speed of light.

[History's Coming To]

This isn't like the speed of light, it is quite possibly the reason for it.

Re:JUST like the speed of light.

[mhwombat]

.... what?

Re:JUST like the speed of light.

[History's Coming To]

There's no good reason why the speed of light is 3x10^8m/s. You can't derive that number from first principles as far as I know. If, however, there's a maximum rate of data transfer or calculation built into the structure of the universe then it would be a very good potential mechanism for explaining why the speed of light is the way it is. I'll have to try some napkin calculations.

Passing the buck

[suso]

Eh, let's let the singularity first, then we'll let the robots take care of the problem.

Re:Passing the buck

[Killer Orca]

Eh, let's let the singularity first, then we'll let the robots take care of the problem.

You mean us?

Re:Passing the buck

[The_mad_linguist]

But the singularity might accidentally the world.

Re:Passing the buck

[Anonymous Coward]

The answer to the problem is just to slow down our perception of time by transferring into a simulation. Then we can violate any law of physics we want.

No growth can go on forever

[jopet]

and no exponential growth can go on for just a comparatively very short time. This should be self-evident, but for some reason, people seem to ignore that. Especially people who call themselves journalists or economists.

Re:No growth can go on forever

[vertinox]

and no exponential growth can go on for just a comparatively very short time. This should be self-evident, but for some reason, people seem to ignore that. Especially people who call themselves journalists or economists.

As far as we know the expansion of the universe and entropy will go on forever.

I doubt that is what you mean though...

Self contained systems do have limits unless of course they are self recursive and halo-graphic.

Like fractals and information...

Economies and ecosystems are not.

Re:No growth can go on forever

[Anonymous Coward]

Entropy by it's very definition will not go on "forever". It'll go on until it burns out, and then it will end. Really, that should be fairly straight forward. Additionally, the universe may or may not expand forever, but expansion isn't adding any new materials or energy to the universe, so in fact, the expansion is adding to entropy.

Re:No growth can go on forever

[mcrbids]

Economies and ecosystems are not.

Economies have to do with the transfer of money. Money is an abstract concept used to replace physical goods. Money is rarely found in its original form (precious metals) and even its derivative form (paper currency) is fast disappearing. Money now is nothing more than bits in your bank's computer.

In other words, money is information. So what sets its limits? The stuff that it represents. Which nowadays is most commonly... more information!

Re:No growth can go on forever

[Anonymous Coward]

Actually, the universe will die a heat death.

Re:No growth can go on forever

[H3g3m0n]

Well the exponential growth has been going on since the very first computer systems where made way back before Moore's law even existed, when they where still using vacuum tubes.

Arguably it goes back even further if you consider it to be systems for information processing rather than strictly computer systems. Things like writing, transcribing books, the printing press, mathematics and so on.

The growth will probably stop eventually but it's unlikely to occur before any kind of singularity happens, even if it does the world will be drastically different. The planned 2012 IBM supercomputer should have about enough processing power to emulate a human brain (its not doing that but they have the blue brain project underway), By 2025 a $1000 computer should have that power (consider what the super computers of that time will have). Unless you think Moore's law is going to kick us all in the nuts in the next 15 years we should be well on our way. Traditional Moore's law (as it applies to transistors on silicon) should continue till some time around 2030 (although some earlier limits are as low as 2020 and it might slow down things leading up to the point). This doesn't take into account the dozens of other non-traditional technologies under research that aren't Moore's law relates: memsistors, photonic computing, DNA/quantum computing (only useful for some specific computation but AI might apply), 3D-ICs, carbon nanotubes, graphite, spintronics.

After some kind of singularity (assuming we survive) we have no idea what the limits are, can we make new sub universes where the laws are better optimized for computing? or change the laws in some specific area? Can we use the theory of relativity to speed computation up (ie I leave a computer on the planet and travel at close to the speed of light in a circle until it finishes number crunching, or hopefully some similar system on a chip)? Can we find some ultimate universal loophole for infinite energy/computation? A cpu that works in an infinite number of parallel universes? Maybe we will hit the universal wall, but by that point it won't matter so much.

Re:No growth can go on forever

[TheTurtlesMoves]

"Humans do not understand exponential growth" --Unknown.

Thats doubly true when its with money....

Re:No growth can go on forever

[vertinox]

In other words, money is information. So what sets its limits? The stuff that it represents. Which nowadays is most commonly... more information!

Good point. I suppose if one wanted you could put an infinity symbol in people's balances of their accounts and just say they can't withdraw physical paper money anymore.

Though I don't think that would work for inflation...

Form over function

[Calmiche]

I figure it will be sort of like the netbook war of today. Manufactures will realize that there isn't much of a way to get faster so they will start concentrating on design, reliability and lifespan. It will probably be a golden age in computing.

I'm just waiting for a peta-hertz computer with a 500 exabyte hard-drive able to do universe simulations in real time that will fit in my pocket, go 100 years on a charge and be indestructible.

Re:Form over function

[fuzzyfuzzyfungus]

Just remember, though, that performing any universe simulations that evolve to include copyrighted works will be a capital offense by the time such hardware is available...

Re:Form over function

[sexconker]

Make sure to NOT handle the recursion!

public static void main Simulate(Universe){

    universe Universe = getUniverse();

    for(BigInteger i = BigInteger.valueOf(0); i < Universe.NumOfObjects; i++){
        if(Universe.getObject(i) == THIS)
            bailOut();
        else
            simulateObject(Universe.getObject(i));
    }

    System.exit(42);
}

Re:Form over function

[sexconker]

I went Java to throw in the BigInteger class (big numbers are always fun to play with), and I preferred "System.exit" to "return".

I haven't touches Java in ages though, so the damned first line is all fucked.

Oh well.

Re:Form over function

[pclminion]

I'm just waiting for a peta-hertz computer with a 500 exabyte hard-drive able to do universe simulations in real time that will fit in my pocket

It is impossible to simulate the universe. This is pretty easy to prove. If it was possible, using some device, to simulate the universe, then it is not actually necessary to simulate the universe -- we only need simulate the device which simulates the universe, since the device is necessarily contained within the universe. This should be easier, because the device itself is much smaller than the entire universe.

But if simulating the device which simulates the universe, is equivalent to simulating the universe, then that would mean that the complete set of states which define the universe can actually be represented by some subset of those very same states -- the subset of states which describe the device which is being used to simulate the universe. In other words, the universe is a set such that if you remove some subset of states you end up with the same set again. I hope you can see how this is a logical impossibility.

Re:Form over function

[MrNemesis]

I fully expect humanity's first encounter with extraterrestrial intelligence will quickly escalate into warfare the likes we've never even dreamed of. Some of the first things they'll receive from us will be the patchy recordings of The Beatles blurted our all over the radio, which they'll have copied and distributed to their scientist caste (the ones with the silly aerials on their heads) for analysis.

As soon as the RIAAKU (Recording Industry Association of All the Known Universe) finds out that their intellectual property has been rebroadcast by the aliens own radio transmitters to a potential audience of billions of trillions (there are alot of stars in the galaxy), the alien race will be sued into oblivion. If they don't die from being forced to forfeit the right to gravity on their planet (due to it being an aiding agent to piracy) they'll be killed by fallout when Britney, aka Space Smallpox, makes its way to their homeworld.

Re:Form over function

[SleazyRidr]

Well, you could simulate the universe, but you'd have a couple of serious assumptions built in.

Re:Form over function

[Parasome]

Only if all states would be necessary for a simulation. Non-random information (needed for a simulation) can generally be represented using less (compressed) information, and the universe is apparently not "random", but exhibits a great deal of structure. I do not mean to say that simulation would be possible, but I think your argument does not prove otherwise.

Re:Form over function

[sabt-pestnu]

A) If you are dealing with infinites, then removing a finite set of substates still leaves an infinite set of states.

B) You don't need to simulate the entire universe. You just need to simulate the portion of the universe that you experience. The Brain-in-a-jar problem. Simulating the "portion you experience" would be indistinguishable from "the entire universe" for a sufficiently adept simulation. From the inside, your tests are, of course, part of the simulation.

Re:Form over function

[Anonymous Coward]

Knowns:

A: Device is finite

B: Device can simulate the universe completely.

C: Device exists within the universe

Conjecture based on postulates:

1) Universe is finite (because device is finite and can simulate the universe)

2) Device is subset of the universe (because it is inside the universe)

3) A subset of the universe is the same size as itself

4) The universe is infinite: this cancels 1, which means one of the "Knowns" is false.

Re:Form over function

[Anonymous Coward]

Even with an infinite set, removing elements from the set changes the set. For instance, the set of "all the integers except 0" is NOT the same set as "all the integers," even though they have the same cardinality.

What is the limit?

[Hatta]

So what is that limit? What units would you express such a limit in? The fundamental unit of information is a bit, what is the fundamental unit of computation? Would you state that rate in "computations per second"? "Computations per second per cm^3"? "computations per second per gram?"

I checked out the pdf of the paper, and didn't see any numerical limit stated, just equations.

Re:What is the limit?

[HTH NE1]

A more practical question: how many bits does my encryption key need now to make brute force cracking impractical for the fastest computer possible in this Universe (i.e probability of finding the key within my remaining lifespan 0.0001% (1 in a million))?

And not involving a system that reduces my lifespan, such as one failed attempt kills me, smart-ass.

Re:What is the limit?

[Idiomatick]

You could move everyone on earth at near light speed while you sit in a pod in space. You will die as a few seconds pass for them and you wont need a very big key, also your life span was not reduced.

Re:What is the limit?

[geekoid]

Speed of light.

Re:What is the limit?

[SchroedingersCat]

2.56 × 10^47 bits per second per gram Ref: Bremermann's limit

Re:What is the limit?

[ijakings]

Computations Per Library of Congress

Re:What is the limit?

[jonbryce]

Given that brute force attacks scale to multiple processors better than just about any other task, I don't think there is a limit.

Re:What is the limit?

[Anonymous Coward]

Easy. One failed attempt kills your information. No shortening of your lifespan involved. :P

Re:What is the limit?

[Gudeldar]

If you use a symmetric key that is the same size as the message (i.e a one time pad) then you can be sure no one will ever crack it without the key.

Re:What is the limit?

[Hatta]

So a bit is a unit of computation as well as a unit of information? How do you determine how many bits you need to do a computation? Lets say, 1 + 1 = 2 (or 10), how many bits is that?

Re:What is the limit?

[SchroedingersCat]

From Wikipedia: "a computer the size of the entire Earth, operating at the Bremermann's limit could perform approximately 10^75 mathematical computations per second. If we assume that a cryptographic key can be tested with only one operation, then a typical 128 bit key could be cracked in 10^37 seconds. However, a 256 bit key (which is already in use in some systems) would take about a minute to crack. Using a 512 bit key would increase the cracking time to 10^71 years, but only halve the speed of encryption."

You see - the system that threatens to reduce your lifespan is a much faster way to acquire that key.

Re:What is the limit?

[SchroedingersCat]

Unit information is a bit, unit of computation is the rate at which the information changes. 1+1 is somewhat complicated example. Let's say you wanted to invert 1 bit. You need a piece of physical matter to encode 1 bit and then being able to measure when the state of that matter changes its representation from 1 to 0. According to fundamental physics you cannot perform that operation faster than Bremermann's limit.

Re:What is the limit?

[doshell]

I would venture that the bits refer to the input size. So 2.56e47 b/s/g means that, if you have a 2.56e47-bit input, an O(n) algorithm (the fastest you can have assuming the algorithm must look at all the input before making a decision) cannot terminate in less than a second in a computer whose mass is less than one gram.

I'm just speculating, of course. My interpretation might be totally wrong.

Re:What is the limit?

[Rhinobird]

I would imagine a unit of computation would have a delta in front of it...to indicate a change

like delta bits per second per gram

Re:What is the limit?

[Anonymous Coward]

What he means is this:

Logical operations can be composed of other operations.

For instance, it's possible to build all the Boolean logical operations (like AND, OR, NOT, etc) out of combinations of the Not-And operator.

In 1980, one of the two physicists here showed that there is one such quantum operation that you can use to construct all other logical operations.

This paper is about the fundamental speed at which you can do that particular quantum operation.

So, "one computation per second" means one application of his quantum operator.

Re:What is the limit?

[Anonymous Coward]

I checked out the pdf of the paper, and didn't see any numerical limit stated, just equations.

That's because the linked PRL paper in the article summary is incorrect. It is two years old, not "published today" like TFA states. The actual PRL paper, published today, can be found right here.

Re:What is the limit?

[physburn]

The paper referenced above is at arXiv, and doesn't give a maximum computer speed per see. It just proves that a quantum running at R computational steps per second, will generate Q = hR^2, of heat, where h is plancks constant. The other limits was that R < 4E/h, where E is the average energy of the system. You might get a maximum computing speed out of this, but only if you have a fundamental limit to how fast you can cool the computer. Not sure where the're fundamental limit come from if not in the above paper.

---

Personally I think, Moore's Law will run out somewhere the early 2020s, and have blogged about what such a computer might be specced as.

Re:What is the limit?

[richard.cs]

The Von Neumann-Landauer limit suggests that each bit of information lost requires ln(2)*kT of energy which is released as heat. Assuming that, as a minimum, it is necessary to flip through all the bits in the key to brute force it then a 512 bit key which requires 2^512 - 1 bit flips corresponding to an energy of 4*10^133 Joules if done at room temperature or around 6*10^121 Joules if done at the coldest temperature yet achieved which is 450 pico Kelvin. Given a universe of mass 1.6*10^55 kg Einstein's relation suggests than if this were entirely converted to energy then only 1.4*10^72 Joules would be available. So even converting the entire universe to energy and using it to run your computer you'd still fall short (by a factor of 10^49) of the energy required to brute force that 512 bit key.

Re:What is the limit?

[corbettw]

From Wikipedia [wikipedia.org]: "a computer the size of the entire Earth, operating at the Bremermann's limit could perform approximately 10^75 mathematical computations per second.

Yes, and all of those calculations would end up along the lines of "6*9=42".

Re:What is the limit?

[khayman80]

Nitpick: that's 10^(-37) seconds, or ~2M Planck times.

Re:What is the limit?

[Anonymous Coward]

then a typical 128 bit key could be cracked in 10^37 seconds.

I think that should be "10 to the minus 37 seconds".

Re:What is the limit?

[chebucto]

[pendant]

computer speed per see

rather,

computer speed per say

[/pendant]

Re:What is the limit?

[DamnStupidElf]

There are presumably a finite number of atoms in the visible universe that we could turn into a computer, so an upper limit on the number of operations per second that we can devote to cracking a key. Assuming no method better than brute force is available (the biggest if, IMO), you just need to find the number of bits that puts the universe into the heat death without breaking it. I've heard that 128 bits is good for a few billion years, and 256 bits will outlast the heat death. Your mileage may vary.

Re:What is the limit?

[Anonymous Coward]

Take a look at this post: http://www.schneier.com/blog/archives/2009/09/the_doghouse_cr.html Basically, you can't brute-force a 256 bit key.

Re:What is the limit?

[Anonymous Coward]

you could maybe argue that in addition to a computation speed, there is an ultimate limit to the number of computers you could make based on the number of particles in the known universe... might add a few zeros to that key though.

Re:What is the limit?

[Mephistro]

Your answer is, per se, another epic fail. Rejoice!

Re:What is the limit?

[blues_shuffle]

" ... a typical 128 bit key could be cracked in 10^37 seconds. However, a 256 bit key (which is already in use in some systems) would take about a minute to crack ... "

Why would a key that is twice as long take 36 orders of magnitude less time to crack?

Re:What is the limit?

[Anonymous Coward]

per se

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

Parent made a typo, you have a more fundamental ignorance.

Re:What is the limit?

[khayman80]

Why would a key that is twice as long take 36 orders of magnitude less time to crack?

Because my point about this being a minor typo was modded "overrated" at the initial score of 2, so it's probably below your threshold.

Re:What is the limit?

[ucblockhead]

Rate computing cycles required to crack a key increases with key size == Exponential

Rate computing power would increase by increasing the number of processors == Linear

Therefore, you can always increase the key size to the point where it is impossible for available computers to crack.

Re:What is the limit?

[HTH NE1]

might add a few zeros to that key though.

Yeah, just make sure they're scattered in the middle of the key and not at the beginning or end, thanks.

Re:What is the limit?

[Anonymous Coward]

...then a typical 128 bit key could be cracked in 10^37 seconds. However, a 256 bit key (which is already in use in some systems) would take about a minute to crack. Using a 512 bit key would increase the cracking time to 10^71 years...

so 128 bits takes longer to crack that 256 bits? anyways, the article you wikimisquoted states the 128 bit cracking time as 10^-37 ... notice the minus sign before 37.

Re:What is the limit?

[Anonymous Coward]

This is simply the time requirement--there is also an energy requirement. I seem to recall reading somewhere that it would take more energy than is released in your average supernova just to put a 256-bit counter through all its states, much less brute-force a 256-bit key.

Re:What is the limit?

[Anonymous Coward]

I believe that is a maximum limit on irreversible computations. That doesn't apply to reversible computations that are used in a quantum computer. This result is a theoretical limit on reversible computations, and is therefore a different result.

Re:What is the limit?

[DriedClexler]

Pointing out that someone was off by 74 orders of magnitude isn't a nitpick :-P

Re:What is the limit?

[Anonymous Coward]

Apparently his wikipedia copy/paste failed to copy whatever character wikipedia uses for negative exponents.

Re:What is the limit?

[Anonymous Coward]

Your correction was very much needed. I was about to post this, and basically assume the whole entry was trash...

128 bit: 10^37 seconds

256 bit: one minute

512 bit: 10^71 years

What's wrong with this picture? (hint: 10^37 seconds is a LOT more than a minute)

Re:What is the limit?

[dylan_-]

Assuming that, as a minimum, it is necessary to flip through all the bits in the key to brute force it then a 512 bit key

Interesting calculation! Minor point: on average you search through half the keyspace in a brute force attack, so use 2^511 rather than 2^512.

Re:What is the limit?

[BuzzSkyline]

That's 'cause i posted the wrong link. This is the right one http://link.aps.org/doi/10.1103/PhysRevLett.103.160502

Re:What is the limit?

[BuzzSkyline]

Here's the correct link http://link.aps.org/doi/10.1103/PhysRevLett.103.160502

Re:What is the limit?

[selven]

The most powerful distributed computing system does 10^16 calculations per second. Multiply by 10^16 and you get 10^32 per second (that is, 2^106), a 128 bit key will last 2^22 seconds (about two months). A 256 bit key will last 2^150 seconds (about 10^37 years). Assuming you intend to survive until proton decay (10^31 years), you got your figure of one in a million exactly.

Re:What is the limit?

[mhwombat]

Given that the paper is about the limitations of a perfect quantum computer, you're going to need to improve your encryption algorithm, not just your key.

Re:What is the limit?

[Anonymous Coward]

a computer the size of the entire Earth

But how much would it take to crack the meaning of life, the universe and everything?

Re:What is the limit?

[Zenaku]

"1+1 is a somewhat complicated example."

I love it. Any discussion in which this statement can be made is a good discussion in my book. :)

Re:What is the limit?

[skeeto]

It comes down to a lack of energy. Flipping a bit takes a minimum amount of energy, kT, where T is the absolute temperature of the system and k is the Boltzman constant. If you were to build a Dyson sphere around the sun and capture all it's energy perfectly an ideal computer using all this energy without loss would only be able to count up to 2^192 in 32 years. As Bruce says in that linked article, "[B]rute-force attacks against 256-bit keys will be infeasible until computers are built from something other than matter and occupy something other than space."

The reason to use a longer key is not to make brute-force attempts take longer, as brute-forcing even 128 bits is never going to be possible, but to cover for potential weaknesses in the cryptographic algorithm that might reduce the strength of the key.

Re:What is the limit?

[Anonymous Coward]

Hmm ... see equation 14. It gives the energy E dissipated per operation in terms of the number of operations per second R. It's a linear relation - so the rate at which energy is dissipated as heat, Q, is proportional to R^2 as they say in the next sentence. (They typo 'heat' as 'head'.)

So if you've got a certain fixed power supply (and hence amount of power that you can dissipate as heat), then the maximum rate at which you can perform operations is proportional to the square root of that amount of power. So the units of the limit are (operations/second)/sqrt(Joule).

Note that if your minimum power consumption goes up as the square of the computation rate, you should be able to save power by running several computers in parallel. I don't know if this is relevant here, though - perhaps reversible computing requires that the entire job must be executed on a single processor (or a set of entangled processors, which may count as a single processor under this treatment).

Re:What is the limit?

[sabt-pestnu]

Conversely, your key could be cracked Between the Strokes of Night.

Re:What is the limit?

[Idiomatick]

So it is safe to say 512bits should be enough for anyone? haha

Re:What is the limit?

[Anonymous Coward]

Building 2^256 computers to get the correct solution one of them got with just one try?
That sounds like a bit hard to me. You are going to need a lot of silicon.
Because addressing them all would require a large bus, each computer would have a unique value and then you would send your encrypted data to them all via radio(save money for the intergalactic transmitter). When one of the computers found it had the right answer it would report back the key. Great I've got a great protocol I should patent it.
Oops but there is another problem, 2^256 computers don't quite fit your basement. How many light years would there be between your computer and each of the bruteforcers?

Re:What is the limit?

[mhwombat]

I can give you my understanding from reading it. The rate R they use is quantum operations per second. I believe that there are only equations in the paper because you have to assume some input numbers in order to get ouput numbers. I don't know what assumptions they made to get the numbers in the article.

I think Equation 8 is a good example because it's about qbits and quantum operations. I'm afraid I don't know a good way to get equations into /., but it's

Q = h R^2 e / 2

In the paper that e is actually an epsilon. Here Q is the heat dissipated per unit of time, R is the number of quantum operations per unit of time, h is Planck's constant and e is the probability of error (i.e. errors per operation).

Hence if your quantum computer is error-free, it doesn't interact with its environment, doesn't dissipate heat, and this "limit" is no problem. But otherwise, the heat dissipated goes up quadratically and it's going to impose a limit. How big a limit depends how much heat you can cope with and what your error rate is. It doesn't seem to give some kind of fundamental hard limit to computing speed analogous to the speed of light, as the summary suggests.

Are there really limits?

[wb8wsf]

Though there might be a limit on how fast a computation can go, I would think that
parallel systems will boost that far beyond whatever limit there may be. If we crash
into a boundary, multiple systems--or hundreds of thousands of them--will continue
the upward trend.

      I suppose there is also the question of whether 10^16 more computing power "ought
to be enough for anybody". ;-)

Re:Are there really limits?

[Anonymous Coward]

Parallel computing won't help.
There's a limit to how fast your compute subsystems can exchange data as well.

Re:Are there really limits?

[stevelinton]

The limit is on the amount of computation you can do per gram per second. Unless your computer was VERY VERY dense and compact (close to being a black hole, in fact) then it would have to be parallel to achieve this limit.

No worry, we'll never get close

[onyxruby]

With the overhead of DRM and other measures that suck cpu cycles on a heavy basis, we'll never get close to the limit. Can we get a new Moore's law, one that includes the DRM tax on our CPU cycles?

Re:No worry, we'll never get close

[sqlrob]

There's a corollary somewhere that the speed of software halves every 18 months, it's been around for years.

Re:No worry, we'll never get close

[tepples]

There's a corollary somewhere that the speed of software halves every 18 months, it's been around for years.

"Somewhere" is probably some article that mentioned Wirth's law, possibly under the name "Gates' law" or "Page's law" or "Reiser's law".

Proprietary journals

[musides]

What an intriguing idea. The article really whet my curiosity. Then to find, as is all too common for scientific journals, that I can't read the damned paper itself without "buying" it. How anti-climatic.

Does anyone have further insight into their ideas?

Re:Proprietary journals

[Anonymous Coward]

"What an intriguing idea. The article really whet my curiosity. Then to find, as is all too common for scientific journals, that I can't read the damned paper itself without "buying" it. How anti-climatic."

Complain that it was paid for with your tax dollars and demand a free copy.

"Does anyone have further insight into their ideas?"

Hard limit+hard limit= hard conclusion.

Re:Proprietary journals

[Achromatic1978]

Complain that it was paid for with your tax dollars and demand a free copy.

To which the typical response will be "only partially, that's why we have these fees, to recoup the cost of the investment. Ergo, as your tax dollars didn't pay for the entirety of the cost of publication, you're not entitled to it gratis".

Re:Proprietary journals

[iris-n]

You should pay more attention to arXiv. Most research published these days is there. This article is. Which makes me wonder, why have they linked to PRL and not arXiv.

Re:Proprietary journals

[tqft]

arxiv is your friend

http://search.arxiv.org:8081/paper.jsp?r=0905.3417&qid=null&qs=Lev+B.+Levitin+and+Tommaso+Toffoli&byDate=1

most released papers have a pre-print arxiv version.

Search by lots of different things

Human Brain Anyone ?

[Anonymous Coward]

I wonder if someone would calculate the real capacity of a human brain and compare it to this limit.
That would mean that there are still ways to evolve, I'm assuming we are quite far from the limit,
and all the ideas about computers getting smarter then man will get a new twist. Since the maximum
computational abilities are limited, then the outcome is not as straightforward as most SciFi novels
potray.

Parallel processing...

[argent]

We're already hitting clock speed "brownouts", and using parallel processing to get around them. To really tell where the limits are you need to look at how small you can make a processor (best case, something like say one bit per Planck length) and how much latency you can afford as information propagates from processor to processor at the speed of light or less.

Not as scary as it sounds

[Stratoukos]

The summary makes it sound like in 80 years there will be no room for improvement and everyone will just have to make do with what they have.

If I understand correctly, the limit is about the performance/volume (performance density?). I imagine that in 80 years most of computational resources will be somehow networked. This means that if I required more processing power than technically possible in a normal computer I could just use someone else's idle processor.

Re:Not as scary as it sounds

[fuzzyfuzzyfungus]

The efficiency of using somebody else's idle processor is necessarily bounded by the speed of light.

The further away your neighbor's blob of idle computronium is, the higher the latency you incur by using it.

I can't believe it

[Anonymous Coward]

"At the current Moore's Law pace, computational speeds will hit the wall in 75 to 80 years"

So soon? I'm dumping all my tech stocks! So computer chip speeds will max out at the speed of a room full of human brains. Outside of attempting to model the Universe what do you need that much power for? Not graphics, games or robotics. Realtime sequencing DNA would be a breeze long before you'd max the limit so what practical use are we missing out on?

Re:I can't believe it

[Anonymous Coward]

So soon? I'm dumping all my tech stocks! So computer chip speeds will max out at the speed of a room full of human brains. Outside of attempting to model the Universe what do you need that much power for? Not graphics, games or robotics. Realtime sequencing DNA would be a breeze long before you'd max the limit so what practical use are we missing out on?

Is this movie really so old that people have forgotten it?

Reminds me of a joke

[jcoy42]

A scientist and an engineer are lead into a room. They are asked to stand on one side. On the opposite side is Treasure (or delicious cake if you please).

They are told that they may have the prize if they can reach it, however they may never go more than half the distance between them and it.

The scientist balks claiming it is obviously impossible as he can NEVER reach the prize and leaves the room. The engineer shrugs, and walks halfway to the prize 10 times or so, says "close enough" and takes it.

So I guess we'll just see, eh?

Re:Reminds me of a joke

[Anonymous Coward]

And a mathematician would stand for a moment, calculate the limit, and then run fullspeed into the wall.

Re:Reminds me of a joke

[Anonymous Coward]

So the tester uses ambiguous language and you use that to try to prove an unrelated point?
If by 'you' you meant any part of the body then the engineer moving his hand all the way to the cake disregards the rules.
Here, here's an example http://xkcd.com/169/

Re:Reminds me of a joke

[bertoelcon]

You need to be thinking with portals.

Re:Reminds me of a joke

[IorDMUX]

Very true.

... though every version I have ever heard of the joke/tale replaced "Treausre" with "attractive woman". This was partially due to the (sadly) male-only content of my graduate-level EE classes.

Re:Reminds me of a joke

[nick_davison]

And were the engineer a hacker, he'd pick up the scientist, carry him half way across the room, set him down and say, "Your turn."

The game changing hackers are the ones who don't listen to the conventional logic of the time and figure out how to wander along a totally different axis that the "experts" hadn't thought of yet.

Look at Wolfenstein/Doom. 3D graphics "weren't possible" on home computers at the time. John Carmack turned it in to a 2D solution and solved it anyway. Perhaps not perfect in every regard but still a hell of a lot better than what anyone else were managing.

Nick's law: at least every 18 months, someone else will declare a limit to Moore's law [and turn out to be wrong].

With our current understanding of transistor science, I'm sure their point is a wonderful one. Problem is, with enough money behind finding the solution, someone'll come up with another axis to wander along that'll continue the advances. But don't feel bad, I'm sure plenty of people thought cart science had reached its theoretical peak and man would never move faster than horses were capable of, too.

Re:Reminds me of a joke

[mrsurb]

He breaks the rules... but gets the yummy yummy cake!

Re:Reminds me of a joke

[Anonymous Coward]

The scientist only makes it halfway out of the room though.

Re:Reminds me of a joke

[Anonymous Coward]

Actually, an accountant was with them too. He walked straight to the cake, and called it a rounding error.

Re:Reminds me of a joke

[Anonymous Coward]

The cake is a lie

Re:Reminds me of a joke

[Hurricane78]

Doesn't this joke originally have a third person? He is a mathematician, and dies inside the room. They find a piece of paper with him, with "Let's first assume I can reach the treasure/cake..." written on it.

Re:Reminds me of a joke

[CubicleView]

The banker was shit out of luck then.

Re:Reminds me of a joke

[CubicleView]

All the Moore's law stuff is presumably just to make the article (which was very light) more interesting or something. This is just an attempt to put a number on the upper limit to the speed of a calculation, it has nothing to do with silicone or any of that. There has to be a finite limit to the speed of a single calculation, adding more processing units will give you a more powerful computer but it won't make it faster than the speed of that one calculation. Presumably they could also dream up with similar theories for the minimum mass required for a calculation. That would probably lead to a figure for the maximum power of a computer, since to make it more powerful you'd have to add more mass. The more mass you add the further information will have to travel. The speed of light would then be a barrier to the theoretical maximum power of a computer, assuming gravity etc wasn't an issue first.

Re:Reminds me of a joke

[Tanktalus]

(As an engineer...)

Nah, that's not breaking the rules. After ten "moves", the eleventh move is simply to reach out and grab the treasure. If you average out his body's movement, you'll find that he has not, actually, traversed farther than half way to the treasure. Only a mathematician would consider the leading edge to be representative of the body, whereas an engineer would consider the centre of gravity to be representative (assume a spherical body... hey, no assumption required!), and thus there'd be no problem in reaching out to grab the treasure as long as his centre of gravity hasn't proceeded more than halfway between his previous location and the treasure. Mind you, if it's very heavy treasure, this may be more difficult.

Re:Reminds me of a joke

[Eil]

I wish I would be around 75-80 years from now, because it would be fun to look back on this research and shake my head in much the same manner people do when they read articles saying that by 1980, the world will be so over-populated that all of mankind will starve to death and become extinct.

It's easy to imagine various hypothetical ways that current technological advances could cease. It's much more difficult to predict ways that technology (and indeed, society) will continue to evolve and progress.

Re:Reminds me of a joke

[Straterra]

From what I hear, the cake is a lie..

Just warning ya!

Re:Reminds me of a joke

[mhwombat]

You only said that because there's cake involved!

Re:Reminds me of a joke

[jcohen]

Modern scientists do not believe in Zeno's paradox. This is an ancient article, but take a look at Adolf Grünbaum's "Modern Science and Zeno's Paradoxes of Motion," circa 1968. It's reprinted in Zeno's Paradoxes, edited by Wesley C. Salmon.

The joke's still funny, but it stereotypes scientists as theory-crazed and impractical -- which I suppose is the typical point of view of an engineer.

Re:Reminds me of a joke

[chewy_fruit_loop]

the cake is a lie

Re:Reminds me of a joke

[dreamchaser]

it has nothing to do with silicone or any of that.

It has nothing to do with fake breasts? That's not a surprise. It really doesn't have anything to do with silicon either.

Re:Reminds me of a joke

[Anonymous Coward]

They can walk up to the cake at the same time...

Re:Reminds me of a joke

[CubicleView]

whoops

Re:Reminds me of a joke

[punkrocher]

This says otherwise. The Mathematician should know better.

Re:Reminds me of a joke

[Anonymous Coward]

...and middle management outsources it to a third world country where they don't have such restrictive treasure approaching laws.

Re:Reminds me of a joke

[Anonymous Coward]

If the treasure is more than a few meters away that won't work. How far can you reach while still keeping your centre of gravity in one spot?

Yeah, except for that quantum mechanics thing

[Anonymous Coward]

Seriously, this may be the currently known limit but I imagine there are more than a few things that will be discovered in the next 80 years.

Besides that, quantum computing will very likely obsolete the way we currently calculate how fast something is.

Electricity cost comes first...

[RyanFenton]

At the current rate of progress, so to speak, no one will be able to afford a computer that runs 10^16 times faster than current systems. Even as a gamer, I'm already leery of buying any of the newer video cards and CPU setups, after reviewing the cost in electricity needed to run them for a year compared to my existing system - they use somewhere around 4 times the electricity!

I can understand fitting more transistors onto a chipset, and more chipsets onto a system, but even with nanotech and similar technologies, I don't see much chance for each transistor to use proportionally less electricity to allow 10^16 more of them to be running at once. You'd have to run a conductance cable to the sun to get that kind of power.

Ryan Fenton

Re:Electricity cost comes first...

[anonymousbob22]

Actually, much of the newer components available are far more efficient than their predecessors in terms of power usage.
Compare Intel's newer processors to the Pentium 4 and you'll see gains in both computing power and power efficiency

Re:Electricity cost comes first...

[Anonymous Coward]

Ah our good friend and notorious troll, Ryan Fenton. I see you're still at your old trolling ways!

Video cards only draw their maximum power when they're being used for gaming. But when idle or doing desktop activities, they draw less power than the lower-power-envelope cards because they're more efficient.

Re:Electricity cost comes first...

[skeeto]

Exactly, we simply won't be able to collect enough energy to do computation at the speed of theoretical limits.

Re:Electricity cost comes first...

[Anonymous Coward]

Running a conducting cable seems like a very good idea! That is, if you want to melt some cable.

Luckily enough the sun already blasts us with more then enough energy on a daily basis, so going through all that hassle just to heat up a cable isn't really necessary.

Re:Electricity cost comes first...

[toddestan]

He's still right though. The latest Core i7's draw up to 130W, which is more than even the hottest of the Pentium 4's. Sure, the Core i7 can get a lot more done than the Pentium 4 in the same amount of time, and since the increase in power usage didn't scale with the increase in computing power, the i7 ends up being more efficient. But take a Pentium 4 system and a Core i7 and run both at 100% and the Core i7 system will draw considerably more power.

Anyone else get the feeling...

[Anonymous Coward]

that the ultimate limit is the processes that the universe itself uses to "compute" its own state? That we can only ever asymptotically approach this limit? Once we hit the limit, our computations cease being simulations and become reality.

Re:Anyone else get the feeling...

[UncleWilly]

no, just you.

Re:Anyone else get the feeling...

[gumpish]

that the ultimate limit is the processes that the universe itself uses to "compute" its own state? That we can only ever asymptotically approach this limit? Once we hit the limit, our computations cease being simulations and become reality.

Lay off the bong hits.

Re:Anyone else get the feeling...

[Rennt]

May I recommend to you Cory Doctorow's "True Names".

Re:Anyone else get the feeling...

[molecular]

that the ultimate limit is the processes that the universe itself uses to "compute" its own state? That we can only ever asymptotically approach this limit? Once we hit the limit, our computations cease being simulations and become reality.

You're correct, proof:
A computer that has reaches the performance of the universe-computer, must necessarily be powerfull enough to simulate itself in real-time, because it is contained within the universe. Since a system can never simulate itself in real-time, no computer within this universe can reach the performance of the universe-computer.

Once our computer reaches the limit, it is the universe-computer and therefore calculates reality.

Re:Anyone else get the feeling...

[takapa]

that the ultimate limit is the processes that the universe itself uses to "compute" its own state? That we can only ever asymptotically approach this limit? Once we hit the limit, our computations cease being simulations and become reality.

Agreed. The 'laws' of the Universe can be said to operate at the smallest measure of x/y/z/planck-time/d5/d6/..... If the Universe has a smallest measure in each of these dimensions then there is a finite amount of time and space in the Universe, therefore a fixed amount of processing power. The Universe is simply playing out its rules at the smallest measure and we are happy byproducts of the process. The Universe itself is the most efficient calculator that is possible in the Universe but the Universe has a finite amount of calculations before the state is forever flat at which point the Universe technically still exists but without any entropy, the Universe can be thought of as dead. That said, the Onmiverse has unlimited processing power but the calculations in those Universes as in our own take the form of realities.

Re:Anyone else get the feeling...

[phision]

And maybe actually the universe is single threaded and nothing happens simultaneously. It is just too fast and we do not notice it. The "parallel" computing will be of no help in this case.

Computing to what end

[belthize]

After reading some of the replies and think about the limit I started wondering about exactly what problems existed that would demand more computational power than 10^16 above what we have now.

I'd be interested in hearing of a problem that can be posited now but can't be solved in a reasonable amount of time (say a few days) with that much computational power. I'm sure there are mathematical oddities or encryption schemes that can chew up all free cycles but it doesn't seem like raw computation is the limiting factor for most problems.

Long before then it's seems I/O bottlenecks are going to be a much bigger issue for any *interesting* problems.

Re:Computing to what end

[Idiomatick]

World recreation through a pure physics sim?

Re:Computing to what end

[Anonymous Coward]

Long before then it's seems I/O bottlenecks are going to be a much bigger issue for any *interesting* problems.

A supercomputer is a device for converting a compute bound problem into an I/O bound problem.

But in all seriousness, there are problems that can use that much computing power. Let's take one for example: You have a disease we don't know how to cure. So we create a simulation of you at a molecular level, and then randomly (or better, heuristically) generate chemicals and simulate their effect. Give it enough cycles and it will find one that cures the disease, no? But you'd better have a whole lot of cycles.

Re:Computing to what end

[Dynedain]

Easy.... CG effects for movies.

From 20 years ago to today, Pixar still takes about 1 hours per frame to render their footage. At 24 frames per second for cinema projection, that's a lot of computation power (around 172,800 hours of rendertime for a 2 hour movie) and they're using cutting-edge equipment. As computers get faster, that hour/frame threshold remains and allows for better quality output. We're still nowhere near high-resolution photorealistic rendering at realtime. Games take an incredible amount of computational and graphic shortcuts, even in something as "real" as Crysis.

Also, take a look at what most supercomputers are used for.... whether, biological, and physics research.

And that new telescope that's been on /. quite bit recently.... generates 30TB of data daily. That's an insane amount of data to process, even if the I/O issues are solved.

Re:Computing to what end

[John Hasler]

> After reading some of the replies and think about the limit I started
> wondering about exactly what problems existed that would demand more
> computational power than 10^16 above what we have now.

1% accurate minute by minute weather forecasts for each cubic kilometer of the Earth's atmosphere a full year in advance.

Re:Computing to what end

[king_nebuchadnezzar]

Try any NP complete problem, this includes traveling salesman graph reduction and many others. lots and lots of problems.

Does this mean no warp drive?

[filesiteguy]

I figure that - even if we find the dilithium crystals - we'd need really fast computers to handle space flights, transporter beams, instant food generators, doors that go "shh!" and warp drive.

I guess it is all just fiction after all.

Re:Does this mean no warp drive?

[Idiomatick]

If you ignore the teleporter and holodeck I'm sure everything in star trek could be run on a decently optimized modern computer.

Re:Does this mean no warp drive?

[linux_geek_germany]

please ship me the replicator then :)

Re:Does this mean no warp drive?

[Idiomatick]

Curses, I knew it was a risk making any firm statements about star trek on /. But damn it /. I'm a computer nerd not a star trek geek.

Re:Does this mean no warp drive?

[marcosdumay]

Maybe, but the existence of FTL transtportation would make that limit moot anyway.

Shannon's law again ?

[Anonymous Coward]

If you believe this then i have a truckload of 33.6 kilobaud modems that will be of use to you.

10^16 times faster?

[gestalt_n_pepper]

Finally, Windows will run fast enough to be useful.

Re:10^16 times faster?

[PRMan]

Finally, Windows will run fast enough to be useful.

Yeah, but Windows 79 is releasing next week, and I hear it's kind of slow on only 500 exabytes of RAM...

Re:10^16 times faster?

[Snarkalicious]

Perhaps. But, unfortunately, this means that all jokes about Crysis are, in point of fact tragic mockeries of all those who own the game and will NEVER be able to play it at max settings.

Nothing interesting in article.

[line-bundle]

I RTFA but there is nothing in the article. Only talk of 75 years...

I remember one way to get an upper limit on frequency is using the equation E=hf, the Planck-Einstein relation. For a given amount of energy you can only get so much frequency. But this was a million years ago in my physics class.

It's not a law!!!

[Anonymous Coward]

Sorry parent for hijacking, but need to troll a bit here...

It's not a law!

Re:It's not a law!!!

[neurophys]

If you want to make a product and you expect to hae it in the marked in two years, you would drop the idea if your product does not fit in the marked included effects of Moore's law. Nobody would make a product that don't comply with the "law". It is driving/regulating the marked.

Pål

PDF on arxiv

[sugarmotor]

Thermodynamic cost of reversible computing
thermo-arxiv
February 1, 2008
Lev B. Levitin and Tommaso Toffoli

http://arxiv.org/pdf/quant-ph/0701237v2

Not sure it is the same as in the Phys. Rev. Lett. 99, 110502 (2007) -- linked from the article -- which is from 2007

Stephan

Re:PDF on arxiv

[BitterOak]

Not sure it is the same as in the Phys. Rev. Lett. 99, 110502 (2007) -- linked from the article -- which is from 2007

The abstract is the same, so it appears to be the same paper.

I haven't read the article all the way through yet. Do the authors take in to account the possibility of optical or quantum computers?

Re:PDF on arxiv

[John.P.Jones]

I don't understand why this is groundbreaking, from looking at the paper it seems that the key result is that the energy cost of a computational step is linearly proportional to the rate of computation, but this result is clearly derived in Feynman's lectures on computation. So they just have different constants that arise due to their considerations of error correction and the like, the essential result isn't new at all.

Re:PDF on arxiv

[tqft]

There is a newer paper I found when looking for this one by the same authors
http://search.arxiv.org:8081/paper.jsp?r=0905.3417&qid=null&qs=Lev+B.+Levitin+and+Tommaso+Toffoli&byDate=1
The fundamental limit on the rate of quantum dynamics: the unified bound is tight
"The question of how fast a quantum state can evolve has attracted a considerable attention in connection with quantum measurement, metrology, and information processing. Since only orthogonal states can be unambiguously distinguished, a transition from a state to an orthogonal one can be taken as the elementary step of a computational process. "

Re:PDF on arxiv

[BuzzSkyline]

Here's the correct link http://link.aps.org/doi/10.1103/PhysRevLett.103.160502

encryption breaking power?

[Anonymous Coward]

If the theoretical max speed of calculations has been achieved, can you now calculate the theoretical minimum times for a single machine to crack certain encryption algorithms?

ie. How safe will files encrypted with today's encryption be in 80 yrs?

Fundamental time unit

[imgod2u]

We've been at roughly ~200ps per circuit operation for quite some time and yet processors are still getting faster. Parallel computation, what a novel idea.

Re:Fundamental time unit

[Anonymous Coward]

Sadly, some algorithms do not lend themselves to parallel execution(like median filters).

Re:Fundamental time unit

[TheTurtlesMoves]

And for some algrothims that need previous results have not got faster. Not everything can be speed up by a || computer.

Distance between objetcs

[Anonymous Coward]

Another problem is solving how to get data from between objects - light can only travel so far between clock cycles, so while a processor may be capable of more and more cycles per second, we are already at the point where unimpeded light cannot travel more than a few inches per clock cycle - so other limitations would be on the size of motherboards, where the memory is located etc.

However such limits can be cheated for a while by using pre emption to get the data where it is needed before it is required.

the limit has been known for the past 50 years

[SchroedingersCat]

It is called Bremermann's limit and it has been known for 50 years: 2.56 × 10^47 bits per second per gram

Re:the limit has been known for the past 50 years

[Anonymous Coward]

mod this up

Too bad spped is just a byproduct

[geekoid]

of Moore's law. Moore's law has to to with the cost of a number of transistor in a given space of silicon.

There are practical limits we are running into that are getting harder and harder to solve.
We are approaching the point where 1 particle of metal per billion and ruin a fab process.
In order to bypass that, we will need to self contained fabs; which would have an even more limited lifecycle the current fabs.

This means the cost of chips could rise dramatically. I don't think many people are going to spend 5K on a home computer anymore.

What is happening is that they are going wide. So more chips but not faster chips; which I think is better anyways.

Re:Too bad spped is just a byproduct

[Yvan256]

The faster we switch to multi-core processors, the better. That way we can "map out" the non-working cores just like non-working sectors/blocks/whatever in flash storage. That would be a perfect way to both reduce waste (no CPUs go in the trash) AND lower prices (since there is no rejects).

Example: let's say we have a 1024-core CPU. You can sell ALL of them with different prices depending on how many cores actually work. Of course a 1024-core CPU would cost a fortune, but at least it wouldn't have to ALSO cover for the loss of the 512 working cores CPU.

No Zen

[sugarmotor]

As usual, Zen is ignored. They don't take into account that when nothing happens that can also be your computation (accuracy -> oo).

Stephan

Constrained Freedom

[DynaSoar]

per TFabstract: "errors that appear as a result of the interaction of the information-carrying system with uncontrolled degrees of freedom must be corrected."

Would not quantum teleportation via entanglement provide a means of distributing computation to include massively parallel? Quantum teleportation would provide a constraint that would redefine the problem by redefining the environment (ie. uncontrolled degrees of freedom). Replace Moore's Law with Bell's Theorem.

And does not quantum computing operate on all possible states, with the answer inherent in the wave function? Spew out the entangled qubits as needed and let them fight it out as a quantum form of Swarm.

If a result can be obtained this way, you may still have a problem with simultaneity -- the answer may arrive "before" the question, making it impossible to decode. However the problem then becomes a limitation of spacetime's ability to pass definitive information, and the limit of computation itself if such exists and/or can be measured in this context becomes moot. Being able to error trace via backtrack is similarly hampered but for the same reason and would still be possible post hoc.

But if a computational system is devised that can operate on such principles, and it is to be used for practical calculations, be aware that any defining of arguments will be restricted to the input end and results for comparison and decision making may not yet be available for such decisions (assuming a reasonable latitude of autonomous action). In which case, make sure you teach it phenomenology *before* putting it to work.

Re:Constrained Freedom

[TheTurtlesMoves]

quantum teleportation needs a classic channel. So you have a speed limit of c. Also you get *zero* fanout and there are other *theoretical* problems let alone practical ones. Even the big proponents of quantum computing do not claim that quantum computers will be made obsolete by them. Event the few algorithms (we have almost none) we have need a substantial part done on a classic computer. You can't beat causality with a classic computer either... So i don't know what you are on about there. Unless we are talking about a QC from a science fiction novel.

However you need to understand quantum mechanics to build a transistor. So you still use quantum physics to build a classic computer.

Re:Constrained Freedom

[DynaSoar]

quantum teleportation needs a classic channel. So you have a speed limit of c. Also you get *zero* fanout and there are other *theoretical* problems let alone practical ones. Even the big proponents of quantum computing do not claim that quantum computers will be made obsolete by them. Event the few algorithms (we have almost none) we have need a substantial part done on a classic computer. You can't beat causality with a classic computer either... So i don't know what you are on about there. Unless we are talking about a QC from a science fiction novel.

However you need to understand quantum mechanics to build a transistor. So you still use quantum physics to build a classic computer.

Your last statement is falsified by the very history of invention of the xistor, thus fiction of a far more useless sort than the science fiction both I and TFauthors engaged in. They and I both project from the known and theoretical to the possible. You attempt to argue against projected possibilities by quoting the present state of the technology and theory, neither of which are sufficient at present. They and I make no pretense to restrict ourselves here to what has been done. Such an argument is hardly an exploration of possibility.

They happen to focus more on the classical. I say the classical might be necessary to set up such a construct, but at least a disproof of their claims if not a functional device can be developed and run.

QT needs a classical channel to set up through. Once entangled entities are separated in 3space that channel is not necessary until decoding and verification. That would provide the proof of whether it worked, and since those results would be sitting and waiting on the classical signal, it would have predated it. The separation of the entangled entity would be done with predetermination of what sort of computation is to be done, and so the results of which can be planned for and tested using far simpler devices than a classical computer, or even with other quantum entities with which the computational entities interact in pre-tested ways. Futhermore, "zero fanout" is an estimate of the natural outcome of generating entangled entities. A series of them can be generated with different base states and the results tested for completeness. The fanout is in the making, and potentially in the interaction, not the evolution of the quantum states of the individuals/pairs.

TFA makes claims about the speed of computation but base their claims solely on classical devices described in terms of entropy as measured at the quantum level. My claim is only that theirs is wrong as a generalization because computation need not be restricted to the classical. The factoring of the number 16 was done entirely within the quantum realm and only tested for post-hoc with a classical device. That design was a convenience, not a necessity, and the results existed prior to measurement. If a computation can be set up across spacetime and reliably tested for by a system operating at c, the computation must progress/communicate faster than c in order to be available by the spacetime limited testing device. Bell's Theorem (actually Bell's own proven violation of his theorem devised in support of the EPR paradox) makes specific predictions that if true would allow this, and they have been so proven.

As an analogy, TFA is like Cayley's early 1800s theorizing regarding surfaces in a fluid stream resulting in predictions that a device might be constructed that via interaction with the atmospheric environment travel without contact with the ground. My suggestions are akin to Tsiolkovsky's 1903 "equations" and work predicting the possibility of flight not constrained by the environment, disproving Cayley's specific predictions due to his self-enforced limitation. I'm fairly certain both of them were hounded in spirit if not fact by the stolidly 'educated' telling them that it can't possibly work because we've seen balloons fly, we know how they work, and they use bouyancy, not li

Re:Constrained Freedom

[FrangoAssado]

QT needs a classical channel to set up through. Once entangled entities are separated in 3space that channel is not necessary until decoding and verification. That would provide the proof of whether it worked, and since those results would be sitting and waiting on the classical signal, it would have predated it.

Sigh... That's not true -- that's not how quantum teleportation works. The classical signal is an integral part of teleportation, it's not just necessary for "decoding and verification" (it seems you're confusing this with Bell's Theorem?).

For enlightenment, this is an outline of how you teleport a qubit:

First, the setup:

1. Start with two maximally entangled qubits (e.g., one of the four Bell states). Call them |a> and |b>
2. Give one of them (|a>) to Alice and the other (|b>) to Bob. They can then travel as far away from each other as desired.

Now, the teleportation. First Alice does this:

1. Come up with or receive a qubit |q>. This is the qubit she wants to send to Bob.
2. She performs some pre-determined unitary transformation in both of her qubits, |a> and |q>. (The unitary transformation depends on the selected entangled qubits in the setup.)
3. She then measures both qubits. The result is a pair of (classical) bits, "a" and "q". This measurements destroys both |a> and |q>, but does something funny (the "spooky action") to Bob's qubit. But notice that Bob's qubit is still not equal to what Alice wanted to send.
4. Alice now sends (via any classical channel) the two bits, "a" and "q" to Bob.

To complete the teleportation, Bob does this:

1. Receive the bits "a" and "q" from Alice.
2. Depending on these bits, he will apply one of four unitary transformations on his qubit |b>. (These depends on the selected entangled qubits in the setup.)
3. After this transformation, |b> is equal to the original value of |q>.

Furthermore, the "zero fanout" is a simple and direct consequence of the no-cloning theorem. It doesn't matter how clever your setup is, you'll never be able to make a duplicate of an indeterminate quantum state. This is a consequence of the fact that quantum mechanics is linear.

Windows Vista

[Anonymous Coward]

So how will we ever run Vista?

You mean mathematician instead of scientist

[g2devi]

The scientist would not give up so easily.

The scientist would simply say that the wave function of the cake already overlaps with his wave function and take the cake.

10^16 times faster than today's fastest machines

[Tumbleweed]

Yeah, until I hit the Turbo(tm) button! 11^16, baby! 11, because that's one more, isn't it?

If by "immediately" there is a limit, then yes.

[erroneus]

It is imaginable that through some of that quantum black magic all possible answers are calculated instantaneously and the correct one selected at the same time and delivered upon query. The bottleneck in that would be the speed in which the question can be presented.

Re:If by "immediately" there is a limit, then yes.

[kalirion]

Unless the same quantum black magic can present all possible questions simultaneously...

Subspace FTL field

[caseih]

So the solution is very obvious. Just put the entire computer in subspace field that creates a pocket of reality where the speed of light is faster (many times faster). Course you then have to have some mechanism for speeding up and slowing down data coming in the ODN conduits. It's been commonly done since the early 24th century. All of these pesky "limits" can be worked around with some fancy level-three diagnostics.

Re:Subspace FTL field

[CodeBuster]

All of these pesky "limits" can be worked around with some fancy level-three diagnostics.

Hmmm...well you know, they could probably double the relative efficiency of the ODN conduits by simply reversing the polarity of the intermix chamber.

Re:Subspace FTL field

[BarMonger]

It's been commonly done since the early 24th century.

Hubert Farnsworth: That's why scientists increased the speed of light in 2208

Re:Subspace FTL field

[sabt-pestnu]

I recall a short story of a "US vs USSR" style chess championship (or "Deep Blue" vs another computer...). Each side put up their best computer for the contest.

One side had an ace in the hole, though... they had developed a field that sped up the passage of time. Set a computer in it, and it could calculate all possible moves from a given position in a reasonable amount of time.

So:

Our heros' computer made an opening move.
The foe's computer, able to calculate all possible moves from that position, resigned.

Yay, we did it again!

[holophrastic]

I hate it when they do that. "this is the limit" really means "based on our current understanding of what's available, this is the limit" -- that's great for the present tense, and it's total garbage for the future tense -- not that english has a future tense, and this is precisely the reason.

But that's fine. 75 years huh? There's a very slim possibility that I'll still be here to point and laugh in their face in 75 years. But there's a virtual guarantee that I be able to point and laugh in their face when their boundary is busted long before then.

Welcome to logical induction, it's an heuristic, it doesn't actually work. Stop basing important things on soft-physics. Try hard-physics for a change. You'll find it much more rewarding.

Re:Yay, we did it again!

[shermo]

Yeah it sucks that English doesn't have a future tense. Hopefully one day it will though. I'll be really happy when that happens.

Re:Yay, we did it again!

[holophrastic]

Actually, I used ot think the same way. But it's knid of nice not being able to speak of the future, because whatever you'd be saying couldn't be evaluated as valid anyway. So this way, it forces the speaker to actually express the desired intent.

In short, it allows me to select from words of meaning, rather than "is". "will", "going", "heading", "might", "may", "can", "could", "should", "would", "is", and many more.

So when I do select the appropriate word, I've said so much more.

That's not to say that my audience has any idea what I've done.

Oh great..

[binaryseraph]

So that means I will be 100 by the time Windows runs smoothly on my machine?

Even when you hit the limit you can add more cpus

[Joe The Dragon]

Even when you hit the limit you can add more cpus / other helper chips.

Right now lot of stuff is started to be coded to use many cores / cups / gpu + cpu.

There Is No Limit

[Anonymous Coward]

After you have built the machine that cannot possibly be made any faster, then you build more and distribute your problems among them.
"Reports of my demise have been greatly exaggerated." - Moore

Re:There Is No Limit

[Anonymous Coward]

...until the computer that distributes the workload also hits the limit.

Re:There Is No Limit

[GargamelSpaceman]

Well this thing is going to have a size, and one end of it is likely going to have to talk to the other end of it in most computations. The fastest that communication can happen is the speed of light.

At a very minimum, you aren't going to get an answer to your question any faster than it takes light to travel from the input port to the farthest point on the machine and back if you require that all sub computations have had an opportunity to interact with each other.

Although this might not be a requirement. You might want all solutions tried and to be notified when the first one succeeds. If the questioner is outside the machine, then they will have much less speedy access to computation power than if they are at the machine's center for answering these questions. The ideal would be to ask the question from the center of a sphere.

For example, you might want to exhaustively search a solution space. You would pose a query to the computer around you with each part trying a part of the space, and not notifying you unless it found the answer. When a part finds an answer you get your reply.

Physicists said we could not exceed 2400 baud too

[Glasswire]

Yup, back in the 80s the physicists said it would be physically impossible to provide switching and encoding which would allow phone line communication to exceed 2400 baud in modems. Yet before we gave up on phone lines, the modem builders were giving us 56,000 baud connections.

The theory people arrive a bit late to the party

[DarkOx]

The physics folks might have worked out some interesting details here but that's all it is interesting. The engineers have already moved on. Its not about getting smaller and going faster has largely past the point of diminishing returns already. There are few applications the digital logic we have today can't perform within time constraints. Even our jet fighters are practically flying themselves. In fact our computing machines are so fast we starting to struggle justifying their applications on anyone task not because they are to expensive this time but because they are so fast that their just idle most of the time anyway. Virtualization is more or less going back to time sharing without the pain. Its about doing more at the same time now, hence all the milti-core chips.

Implications for Ultimate Limits

[Vector Meson]

The article is from 2007 but I guess it's news to most nerds.
Arxiv link: http://arxiv.org/abs/quant-ph/0701237

Now we just need to figure out if this has any impact on Ultimate Physical Limits of Computation as linked to in LWN: http://lwn.net/Articles/286233/

The fastest computer physically possible...

[dingen]

... imagine a Beowulf cluster of those!

Limit not well-defined

[curril]

What the article doesn't appear to take into account is the difference between quantum computing and conventional computing. A quantum computer doesn't carry out conventional binary operations so comparing the two is tricky. Doubling the number of q-bits a quantum computer can process effectively squares the number of equivalent binary operations it can carry out in one computation, but quantum computers are limited in what kinds of binary operations they can compute simultaneously. So saying that a there is a fundamental limit on the number quantum operations per second doesn't necessarily give you a meaningful limit on the number of binary operations per second. You need more information, such as the number of linked q-bits and the types of binary operations being performed.

Re:Physicists said we could not exceed 2400 baud t

[Gothmolly]

You mean bps connections - the modulation on a V.90 modem needs far less 'baud' to get that many bits/sec down the wire.

Re:Physicists said we could not exceed 2400 baud t

[Dynedain]

We haven't given up on phone lines.... DSL gets a hell of a lot faster than 56kbps

Nothing a watercooling block cant fix

[Anonymous Coward]

Yes, 10^16 times faster than todays machine, but can I overclock it?

Re: Pesky limits

[roguegramma]

On the other hand, I suggest we put all who suggest the speed of light is so easy vanquished into a superspace bubble, where light moves 1000x slower - then the guys in the bubble will think that the computer outside of the bubble runs 1000x faster. And as a bonus, my idea is admissible by known laws of physics.

What ABOUT Parallelism?

[Nakoruru]

What about parallelism?

This made me facepalm. What about it? These physicist used the laws of thermodynamics to establish a fundamental limit to how much usable information can be inserted or extracted from a volume of space. What does parallelism have to do with that at all? What indeed?

See how easy it is to ask rhetorical questions?

 

Re:What ABOUT Parallelism?

[Anonymous Coward]

See how easy it is to ask rhetorical questions?

How could I miss it?

Re:What ABOUT Parallelism?

[Anonymous Coward]

And yet you continue to post here....over and over. You are such a sad little man. I wonder what Karla thinks of you?

Re:Fundamental mass unit

[Anonymous Coward]

Ahem. The limit is expressed in calculations per second per gram.

That means you can go in parallel all you want, but for the next few millennia, computers will still be bounded by the mass available in our solar system. Obviously at some point you'd have to make the planet uninhabitable to humans to gain any more compute speed.

Maybe a million years from now we'll send a compu-forming (as opposed to terra-forming) mission to Alpha Centauri and start turning it into a new supercomputer with an ~8 year ping time from Earth. ;)

Predictions just outside their lfietime = bull

[syousef]

So we'll have to wait another 75 years before management lets us focus on application efficiency instead of throwing hardware at the performance problems? Sigh...

You're going to place any stock in a prediction made by someone that will only come to fruition just outside their lifetime? Such predictions belong in a tent with a crystal ball.

Computers will soon reach the Amiga Factor

[Orion Blastar]

in which more than one processor will be used for various things like graphics, sound, I/O, memory management etc.

This will be done because they cannot make CPUs any faster clockwise or instructions per second wise, so they will have to use co-processors to work with the main CPU to free up the main CPU, and thus speed up the system.

We might even see parallel processing PCs with more than one CPU chip (not multi-core but more than one CPU) in order to run them in parallel to handle more things.

Some day we will reach quantum computing, fiber optic motherboards, fiber optic RAM, and then we will find way to speed things up more, but until we do, there will be physical limits on what a computer can do speedwise.

I call it the Amiga Factor as the Commodore Amiga used the 68000 running at a slower 7.14Mhz speed, but used co-processors to speed things up and take the tasks off the main CPU. We sort of have that a bit now with modern Macs and PCs as Video Cards have GPUs and some GPUs are built into the CPUs now, and each Sound Card has a processor of sorts. So basically modern systems have evolved into what the Amiga might have been had it used Intel chips instead of Motorola chips.

Two things

[RingDev]

1) If you don't approach science as a whole, from the angle of challenging expectations, you're doing it wrong. We don't prove that theories are "right", we fail to disprove them. So if you find the concept of disproving theories to be personally insulting, you have no business in a lab.

2) Given the attitude you've shown in this thread you appear to have the interpersonal skills of a Hymenoepimecis argyraphaga wasp. If you behave so improperly when not behind a computer, I would venture a guess that you are all but un-employable, regardless of how intelligent you feel you are. If you are gainfully employed, I would appreciate it if you could conduct yourself in a professional manor when participating in a public debate.

-Rick

Re:Two things

[iris-n]

Even though he can't talk, he's right. You have to have some healthy skepticism, but at some point it just becomes stupid.

Can you honestly conceive of "technological advances" that would make FTL communication possible?

Or some engine more efficient than Carnot's cycle?

Or a computer that could compute all the evolution of the universe in a second?

There are things that just don't make sense. These things are so fundamental that to give up on them you would have to give up all of modern physics, and any hope of being able to correctly describe nature.

Re:Two things

[Anonymous Coward]

Yeah, but who's going to go through all the trouble to build a professional manor, just so that cretin can conduct himself in it?

Re:Two things

[shutdown -p now]

Humanity has given up all of its "physics" (or rather what came for it) many times in the past. Some of it (like several Ancient Greeks' theories) was in fact rooted in scientific method. What makes you think that the current iteration is the final one?

In fact, we have very good reasons to believe it's not final, until we get a self-consistent theory of everything with no black spots in it. So far we have general relativity and QM which are for all intents and purposes incompatible, and yet both are accurately describing our observations. This means that our present understanding of the world has some pretty fundamental flaws in it, and is no less an approximation of reality than Newtonian physics was in its time - just a lot more accurate.

Until that question is resolved, talking about impossibility of FTL is rather premature. So far as I know, there are a few complex and exotic, but nonetheless consistent theories of everything - so far unprovable - that would enable something that could be reasonably deemed FTL. They may be unlikely, and we can't yet prove them right, but we can't prove them wrong either, so don't rush it.

Re:Two things

[bkr1_2k]

We can theorize ways to travel faster than light but not to communicate faster than light? Really? Those two don't mesh, for me. Obviously if there are ways we can travel faster than light we'll be able to communicate at such speeds as well.

Re:Two things

[selven]

That third one is the only one that is truly impossible. A computer cannot calculate the universe faster that the universe does because the computer is part of the universe and thus has to out-calculate itself.

Re:Two things

[iris-n]

Ancient Greek's theories rooted in the scientific method? That's news to me. Could you pinpoint one?

You miss the point. No physicist thinks that the Standard Model (or GR) is the final theory; many details of it are open to question, and they'll probably be disproved someday.

But there are fundamental principles that are beyond specific theories. We call them meta-theories. That is, a framework that any theory has to obey to be taken seriously.

Take FTL communication. Its impossibility is implied by the causality principle. And if you've given up causality, you might as well give up physics. Related to it is locality. No theory that allows me to change things in Andromeda while messing with things here can make any sense (quantum non-locality is local).

Or NP problems. Deutsch has a beautiful paper in which he uses time travel to devise an algorithm to solve NP problems in polynomial time. He then argues that what this proves is that time travel is impossible. For him (and I agree), the hardness of NP-complete problems is a fundamental property of nature, and a theory that violates it is just plain wrong.

There's another paper, whose author I forgot, in which is shown that if you add a non-linear term to Schrödinger's equation you can violate Heisenberg's uncertainty principle and the second law of thermodynamics (among other things). Again, what this proves is that quantum mechanics is definitely linear.

Re:Two things

[Rennt]

But the argument here is that we will hit hard limit much sooner then the theoretical limit, not the other way around.

To use your FTL example, not only do I acknowledge the theoretic limit of C, I don't believe we could even achieve HALF that. That IS healthy scepticism.

Re:Two things

[iris-n]

If you think so, you ought to study more physics. All three are equally impossible. But the second one really offends my aesthetics instinct.

Re:Two things

[wwfarch]

But what if we found a way to create extra universes and extract information from them? There are ideas about creat big bangs, etc... so we can't really say this is impossible at this point. Personally, I think it probably is but we can't prove that right now.

Re:Two things

[SleazyRidr]

If we're either old enough, or have studied enough history, we've heard these sorts of arguments many times before. I propose SleazyRidr's law, that the rate at which humanity declares that no more advances are possible doubles every time such a claim is made.

Re:Two things

[Anonymous Coward]

Can you honestly conceive of "technological advances" that would make FTL communication possible?

Well, yes. Yes I can conceive it.

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

Re:Two things

[kalirion]

Can you honestly conceive of "technological advances" that would make FTL communication possible?

Yes, and I'm not the only one. There are legitimate scientists who believe wormholes are possible. Also there may yet be a way to use this whole quantum entanglement business.

Or a computer that could compute all the evolution of the universe in a second?

Sure, as long as the computer exists outside of the universe in question (and therefore outside its spacetime.)

Re:Two things

[TheTyrannyOfForcedRe]

You expect "professional" behavior on Slashdot? Are you new here?

Re:Two things

[selven]

The first two are physically impossible (ie. if new physics are discovered, they may stop being impossible). The third is mathematically impossible, and you can't have different laws of math - 2 + 2 = 4 holds true in any universe.

Re:Two things

[Anonymous Coward]

I'll take that on.

1. Yes, I can conceive of "technological advances" that would make FTL communication possible. Given a stable wormhole, this could be achievable. I see no reason to believe that stable wormholes are, perforce, impossible.

2. Perhaps you mean a Carnot heat engine... which operates using Carnot's cycle. I'll grant you that I *personally* don't have a good faith belief that "technology" is going to find a useful workaround to the laws of thermodynamics.

3. Absolutely. I see no reason that a significantly advanced simulation cannot model (with a reasonable degree of detail/fidelity... planck scale, say) a Universe at an arbitrarily fast speed.

Re:Two things

[Anonymous Coward]

I would appreciate it if you could conduct yourself in a professional manor

This seems to be conditional on whether his place of employment is located in a large house.

Re:Two things

[alexo]

Yes, I can conceive of "technological advances" that would make FTL communication possible. Given a stable wormhole, this could be achievable.

John Crichton, is that you?

Re:Two things

[marcosdumay]

Still, the second one is the weakest.

Re:Two things

[iris-n]

I wouldn't even answer to this, but since it's modded insightful I guess its a common doubt.

Of course FTL travel implies FTL communication. And vice versa. There are monstrosities like wormholes that are seemingly allowed by general relativity, but they always fail to deliver when you examine more closely. The Universe in a Nutshell explains some guesses (only guesses, since we don't have quantum gravity) of what quantum mechanics would do in face of such a time machine.

But the convincing argument for me is that FTL => time travel => (P == NP). Since P != NP, FTL and time travel are impossible.

That's not denying advancements. There are mistakes in the current physics, and there's still a lot of things to discover. But there are fundamental principals that I'm pretty confident we've got right.

Re:Two things

[bkr1_2k]

Fundamental principles like the earth being the center of the solar system?

Or like earth being flat?

Or like ...

Plenty of options for understanding beyond being confident in things we've got right. We only assume they're right because they haven't been proven wrong, yet. Others already pointed that out.

Re:Two things

[stonecypher]

Hi, 1993 calling. We've had FTL information transfer since the first co-vibrated Bose Einstein condensate. Since your knowledge of science appears to be based largely in science fiction, you may believe you know what this is under the name "ansible".

FTL communication was laboratory demonstrated sixteen years ago. Stop telling yourself that you know things you don't actually know.

Re:Two things

[alexo]

Hi, 1993 calling. We've had FTL information transfer since the first co-vibrated Bose Einstein condensate. Since your knowledge of science appears to be based largely in science fiction, you may believe you know what this is under the name "ansible".

FTL communication was laboratory demonstrated sixteen years ago. Stop telling yourself that you know things you don't actually know.

Oh, hello stonecypher. I thought I recognized your style.

The analysis you constructed from a single off-handed remark on "stable wormholes" is quite amusing.
Do continue.

How many ways is TFA wrong..let me count them.

[Anonymous Coward]

"a perfect quantum computer spits out ten quadrillion more operations each second than today's fastest processors"

Its pointless to compare quantum processor operations with todays CPUs.
CPUs operate in terms of single operations with fixed complexity per second. Quantum CPUs can perform single search operations of complexity 2^qbits per cycle.

We don't know enough about quantum phyics to know if its even possible to EVER build useful quantum computers let alone this crazy notion of going around peddling theoretical maximums.

The information propogation limit "C" is roughly 1 FT per nanosecond. Even with three dimensional component stacking and mystical use of single atoms as transisters and you've already more than made up for any practical information limit based on thermal noise.

Re:Physicists said we could not exceed 2400 baud t

[John Hasler]

> Yup, back in the 80s the physicists said it would be physically impossible to
> provide switching and encoding which would allow phone line communication to
> exceed 2400 baud in modems.

Let's see a citation.

Re: Channeling Yogi

[maxume]

There isn't any letting the singularity, if it is going to happen, it will do it on its own, and if it isn't going to happen, it will do that on its own too.

Re:Physicists said we could not exceed 2400 baud t

[Anonymous Coward]

References please? To which "physicists" do you refer? And what were their assumptions that led to their conclusions?

The data speed of phone lines in the 1980s would be more of a concern for engineers than physicists, with all due respect to both professions.

Frankly it sounds to me like you have quoted something horribly out of context.

Amaze us

[LionKimbro]

Yeah, but the people who propagated Moore's law said we'd see things like 1000 GHz computers.

Even if you multiple 3 GHz by 4 (four processors,) you get a measly 12 GHz.

The subjective experience of going from 4 MHz to 8 MHz to 16 MHz to 44 MHz to 100+MHz was extraordinary, and the effects showed with everything.

But do I personally notice much difference on a one-core, duo-core, or quad-core?

I can hardly tell these days, and could care less.

Everything amazing these days is user interface, collaboratively developed rich data sets, disk drive volumes and bandwidths, algorithms, software.

"Oh look, I have 128 cores on my chip." "Can I do anything better with it?" "..."

When we have real-time speech recognition, -- that'll be something really great, really amazing. That can use more processors. But what do we see today that we didn't see 4 or 5 years ago, that relies on the faster computer?

I feel short-sighted-- tell me reasons to be excited about faster computers; Give me something to look forward to.

Re:Amaze us

[Cal27]

You'll be able to run Office and watch a flash video in Firefox...
At the same time.

Re:Physicists said we could not exceed 2400 baud t

[TaliesinWI]

No they didn't. Baud=bitrate only in 110/300 bps era modems. 9600bps (V.32) modems were at 2400 baud but using 4 bits per symbol. Even in the post-1990 modems with trellis modulation the baudrate never cracked 3,429 but with V.34bis we were at 33.6kbps. That was the absolute maximum on an analog-only phone line. Anything past that (V.90/V.92) was one-directional PCM which you could only get away with because modern POTS lines are carried on a digital infrastructure.

Actually

[langelgjm]

Actually, if you could demonstrate that your simulation independently created a work identical to another copyrighted work, the evolved work would not be a violation. In fact, you'd have the right to distribute "your" work, while the other person retained copyright on "their" work. There's precedent for this, but I'm not going to look up the case right now.

Of course the chances of that happening are pretty low... and any attempts to "filter" through all the random crap with the intent of searching for an evolved work that matches a current work would undermine the legal argument.

And in any case, the lobbyists will probably write a provision explicitly targeting this when the time comes :-)

Re:Actually

[fuzzyfuzzyfungus]

I wonder what flavor of "demonstrate" would suffice...

For instance, it is pretty trivially evident that the set of all natural numbers, each one expressed in binary notation, is not under copyright; but nevertheless includes .wav versions(as well as .midi versions, .mp3 versions, and PDFs of the musical notation) of every song ever written(and all the covers thereof).

I'm guessing, though, that this assertion(no matter how clearly proven) would not hold much sway in court.

Re:Physicists said we could not exceed 2400 baud t

[Anonymous Coward]

v.34 is the last standard I'm aware of that didn't require anything special to operate.

It was less than 3600 baud at highest speed. The actual throughput was 33,600 bits per second per channel, thanks to the advanced encoding, of course..

v.32 was 2,400 baud, giving 9,600 bits/sec per channel. The 2,400 "baud" modems were actually 600 baud with 4-bits-per-baud.

v.90, and all of those other semi-digital systems aren't running over a normal phone line anymore, in essence. DSL units are only linking you to the nearest CO. Old school v.32 could connect you to any other point on the PSTN network without special, telco-side hardware.

Also, if you have a digital, T1-based phone system (PRI), each channel is only 64,000 bits-per-second. You're definitely not overcoming THAT with any fancy tricks.

Bah, that Number is just nonsense.

[Timmy D Programmer]

They just picked a number so darn high they would be long dead by the time it's proven wrong. The reality is sooner or later somebody will approach getting more than a boolean out of a single transistor. Of course 'that' probably won't be called a transistor, but you get the point. So you run out of space, and still find a way to get more calculations per second. In other words. Bull Droppings!

Re:Physicists said we could not exceed 2400 baud t

[Anonymous Coward]

Interesting. Can you provide a reference?

Simple Simple Solution

[Nom du Keyboard]

The simple solution at that point is to go parallel.

In other words...

[jonaskoelker]

In other words, quoting a fortune cookie:

Progress means replacing a theory that's wrong by a theory that's more subtly wrong.

New article, old subject

[maXXwell]

This topic has been visited numerous times. A particularly good article on theoretical computational limits appeared in Nature in 2000:

Lloyd, Seth. "Ultimate Physical Limits to Computation". Nature 406, pp. 1047-1053 (31 August 2000)

http://www.nature.com/nature/journal/v406/n6799/full/4061047a0.html

Re:Fundamental mass unit

[jonaskoelker]

Maybe a million years from now we'll send a compu-forming (as opposed to terra-forming) mission to Alpha Centauri and start turning it into a new supercomputer with an ~8 year ping time from Earth. ;)

Be sure to bring a pair of white mice :)

Norton Internet Security

[Anonymous Coward]

Norton Internet Security 2090 will probably eat up any benefit.

Thats why

[doronbc]

You increase the speed of light. According to Futurama, that will happen in 2208, so I think were screwed.

Not the right article

[mesri]

Doesn't anyone read the articles? It says that the article in Phys. Rev. Lett. was published _today_, October 13, 2009. The article that was linked to is two years old and not really relevant. This is the one they're talking about: http://link.aps.org/doi/10.1103/PhysRevLett.103.160502 There's a preprint at: http://arxiv.org/abs/0905.3417 The gist of it is that one can consider a fundamental step of a computation to be the evolution of a quantum system from a state to an orthogonal state (cause if they aren't orthogonal, you're going to get the answer wrong). They figure out the maximum rate at which the system can evolve between orthogonal states, which sets a maximum to the speed of the computation. Turns out that the rate is proportional to the difference in energy of the two states -- which means that you can drive the computation faster by choosing two states that have very different energies. But if you do that, since you need to have a power source driving the system between the two energy levels, you have to spend a lot of energy to keep the rate up. Sort of obvious, but they work out the details with explicit lower bounds for the first time

Pfaugh! Utter rubbish!

[Anonymous Coward]

My - soon-to-be (oh, in a couple of decades or so) - higgs boson (see previous post) based fpu dirac-sea interwave (hence extra ortho-temporal) calculators will immediately avoid all 'answers-that-should-not-be' (metallica?) and place the right answer complex at precisely the same time-space-reality point as the posing of the question. Or sooner, if desired. More advanced versions could supply the answer as the decision to pose it develops. As a shortend-path return-wave artifact, or some other banality. For the terminally lazy, the system could just choose any answer, serve it, and higgs-bosonify the questioning reality into conformity.

Those chaps are obviously in the 'no rocks in space', 'no heavier than air / faster than sound flight'... category. Just ignore them. They eventually go away.

Ghosts or gods

[Anonymous Coward]

This topic needs some theology. We will be truly masters of science when we earthly gods approach the physical limits of computation, as we now know it. What will we need or desire? And, why will we gods care?

Re: Channeling Yogi

[Jeremi]

There isn't any letting the singularity, if it is going to happen, it will do it on its own, and if it isn't going to happen, it will do that on its own too.

Well, somebody is going to have to write that first program that's intelligent enough to write a program that's more intelligent than itself (and also dumb enough to run said program)...

I call BS.

[Kingrames]

K, Here's how I see it.
I can accept that nothing can move faster than the speed of light.

That means that there is a temperature at which nothing can be colder (absolute zero) and a temperature in which nothing can be hotter (the temperature at which all molecular and atomic motion is done at the speed of light).

But I can't see a physical limit to what they're describing, because it's wrong.

Here's why:
I create a holographic projector. it's about the size of a housecat. it creates a holographic image that is the size of 24x30x200 Jamesisawesomes (that's a term for something that makes up the things that make up the things that make up the things that make up the smallest things we know about now). Even though nothing can physically be that small, I've created a hard drive that is virtually that small, allowing for the creation and operation of a ridiculously tiny hard drive that can hold a near-infinite amount of data.

The flaw in their argument is that they assume we won't discover something "smaller than that" which can be used to catapult our technology one more step into the infinite.

Engineers are not Gods

[Secret Rabbit]

While many experts expect technological limits to kick in eventually, engineers always seem to find ways around such roadblocks.

Engineers get around technological limits. They do NOT get around physical ones. This is a limit governed by the laws of Physics. It can't be worked around without new Physics allowing it to. Remember, Engineering is built on these laws.

Core it up

[DoktaDre]

There's no upper limit to the number of cores you can have... or is there?

Re:Core it up

[kalirion]

Pleas break the following problem into a billion parallel chunks to make the computation faster:

2 + 2 = ?

Re:Core it up

[DoktaDre]

Your argument doesn't exactly void the practical benefits of concurrent processing.

Re:Core it up

[kalirion]

I'm not saying there are no practical benefits, I'm saying that some problems cannot be solved faster by adding more cores.

Powerful enough

[jameskojiro]

To build a computer that will fit on my desk that will be able to fake enough of the physics to simulate a person in a virtual world of their own.

I used to think that

[symbolset]

Back in '81 when the US national debt passed a trillion dollars I did some forecasting and estimated that the debt escaped to infinity in 2012. I was really scared about that for a long time.

It's only with the advancements in 64 bit computing technologies that I can see the error I made: advances in our understanding of numbers allow for ever-more absurd extentions of logarithmic growth. The national debt is not even 11 trillion dollars now, and that's just the debt. The government took on over 8 trillion dollars in unfunded obligations last year alone and it would be even more this year even without the healthcare fix. The total unfunded obligations as of the start of this year were 63.8 trillion dollars, or over half a million dollars per household. But those numbers now fit in my iPhone scientific calculator, so it's all good.

Just like the budget is big numbers, the size of components is small numbers. Sometime between now and 80 years from now we'll discover what the component parts of quarks are, and these quarklets will compose our transistors in some way we don't yet understand. Likewise, by then my great grandkids will each owe nearly a trillion dollars of their own and the US debt will be in the septillions, but those numbers will comfortably fit in their 512bit cybernetic math implants so they'll be fine.

That's just progress. It took some getting used to, but it's not scary any more.

Re:I used to think that

[TheTurtlesMoves]

I think you missed the economic meltdown memo. My friends retirement fund went from 100K to less than 7K. The way we expect >3% growth or more (more like 10%) you are guaranteed to have a meltdown once or more in your lifetime, so good luck with retirement. And mark my words, this meltdown hasn't finished yet.

Good.

[Civil_Disobedient]

In 75 years programming languages will be so abstracted for the sake of convenience that even with the extra processing power, programs will still run just as fast as they do today.

I'm looking forward to when we finally hit the wall. Then we'll have no other option but to concentrate on programming efficiencies. Unlike today's asinine "Just throw a faster CPU and a few more gigs of RAM at the problem!"

Programmers who can think in Assembly languages are simply better than today's script-aculous wanna-be's.

Can't wait

[marqs]

So what you say is that i have to wait 80 years before buying the ultimate computer without feeling that there is just some new platform around the corner that will make my $2000 gaming rig look like a pocket calculator?

Just a thought

[marqs]

What if we build the ultimate computer to answer the question of "Life, the universe and everything".
What if said computer ponders the question for seven and a half million years, and then comes up with an answer close to but not quite exactly 42.

Then we know for sure that the CPU has it's own Pentium bug.

Quantum limits

[jandersen]

So, although the article doesn't state so, this is about yet another application of Heisenberg's uncertainty principle - a principle that was inspired by the recognition that particles are also waves in some sense; the uncertainty arises because there is a limit to how much detail you can see when you observe by irradiating your target with particles of any given wave-length. IOW, it doesn't really say that "there is nothing to observe", but rather that there is a limit to the method of observation used.

This of course casts an entirely different light on the validity of absolute statements about the nature of the world on the very small scale; if one could find a method of observation, that wasn't subject to the limits in our current methods, we could improve significantly on Heisenberg's uncertainty. The very fact that QCD seems to work so well, suggests that there are details to be found beyond the 'Heisenberg limit'.

So, it is not entirely impossible that we will find a way round that one.

There is also limit to that

[abies]

There is a sf book which explores this concept in more detail.

http://en.wikipedia.org/wiki/Perfekcyjna_niedoskona%C5%82o%C5%9B%C4%87

Idea is that you create Ultimate Computer which is best thing you can get with current laws of physic. Then, you start to create different universes (or 'inclusions' as named in the book) with 'better' laws of physics and build better computers there (and 'outsource' the computation). At some point you will reach Ultimate Inclusion (best combination of laws of physics for the best computer). Fortunately, around that point, you are supposed to evolve enough to not care anymore...

Thanks

[Anonymous Coward]

I love that quote, did you receive it personally in a fortune cookie?

Hmmm, "The net" seems to sometimes classify it as a quote from Stephen Hawking. Of course, the Google Books hit which actually gives a bibliographical reference to it, doesn't let me look at the page with the bibliographical info. Anyone know if it's a real quote from him?

Re:Thanks

[jonaskoelker]

did you receive it personally in a fortune cookie?

Only to the extent fortune(6) knows me personally ;-)

Lem predicted this one too

[macson_g]

Stanisaw Lem in his novel "Fiasco" (1987) described concept of "ultimate computer" - computer that is constrained only by physical limits - like Planck constant and speed of light.

And then, five seconds later

[Lord Bitman]

Five seconds later, an equation describing the universe as fractal and infinitely divisible surfaces, and Moore's law continues.

Re:The theory people arrive a bit late to the part

[TheTurtlesMoves]

I just let some simulation software run for 2 months on a 250 core cluster. This uses many simplifications so that it wouldn't take years to run. Just because your PC is sitting idle all the time is not the same as all computers are idle. We are just about to spend another 150K EU on another cluster. We need faster cheaper machines... (that use less power... cooling and power connections are eating our budgets)

Re: Channeling Yogi

[maxume]

Well sure, but things aren't likely to get to the point where that program becomes possible and still have 'somebody' be limited to a few people, so it will become increasing difficult to stop everyone capable of creating it, and then, if things don't reach that point, you don't have to stop anyone.

Windows 2075 ViXtaPro OuterSpace Editon

[Anonymous Coward]

will fully exploit your computing power with the new Aeno eye catching GUI accelerator technology

Wrong Link in the Story!!!

[BuzzSkyline]

My bad, the link to the correct paper is this http://link.aps.org/doi/10.1103/PhysRevLett.103.160502 Sorry kids. Buzz

Never, ever say never!

[Anonymous Coward]

"Who'd want a computer at home?" - DEC CEO.
"Travelling faster than 5 mph will kill you!" - Anti-loco nutters 18th C.
"Splitting the atom and making enough energy to trigger a bomb is impossible." - Albert Einstein

Never, ever say never!

Re:PDF on arxiv - same basic content

[mhwombat]

They are the same in essence but not identical. The PRL version appears to have been edited for format and brevity: section headings removed, equations placed in-line etc. There are also a few more material changes such as rearranged paragraphs.

The basic content and equations seem to be all the same so far as I could see.

The PRL version seems more recent, but the arxiv version is actually more readable IMO as it takes up half a page more room and is split into clearly-titled sections.

Makes sense...

[hesaigo999ca]

The smallest distance, with the fastest bus technology times the amount of cpus involved will be the factor in this, and I do think at some point as the article states, we will be able to get no smaller cpus, or no smaller serial bus etc.. there is a limit to the size before you just cant squeeze anything else from the getup. I know we are nowhere near yet, however...when it does happen, people will probably all have 1000cpu computers at home, and we will all be cluster networked for our downtime for some government project, and
everybody will be driving flying cars.

Correct article link

[Anonymous Coward]

I'm not sure if someone already corrected this omission, but the linked article originally provided in this story is not the correct one. (It is an old article from 2007 that relates the minimum energy dissipation per step to the square of the rate of computation and also derives a generalized Clausius principle.) I believe the correct article to refer to is the following:

http://arxiv.org/abs/0905.3417

Re:Correct article link

[BuzzSkyline]

This is it http://link.aps.org/doi/10.1103/PhysRevLett.103.160502

What about windows

[namoom]

if we hit a computational limit how will windows ever survive new releases

Re:Even when you hit the limit you can add more cp

[Anonymous Coward]

Except those additional cores are also governed by the limit, as well as the transmission of data between them.

Wow that's fast...

[ZERO1ZERO]

If the physicists are right, though, no technology could ever beat the ultimate limit they've calculated which is about 10^16 times faster than today's fastest machines

...but can it run Vista?

Time Cube guy, is that you?

[sean.peters]

I'd like to subscribe to your newsletter.

What's insightful about this?

[sean.peters]

Who said anything about whether we were "meant" to compute infinitely fast (whatever that even means)? This article is about whether it's possible. The answer, unsurprisingly, would appear to be no.

Re:What's insightful about this?

[theshowmecanuck]

The point is, that many people want to set limits because they are not capable of believing there is 'more' in the world. Those that do believe are the ones who actually come up with the truly novel innovations and discoveries. Someone out there has deemed that they have discovered the limit of something. They are like the crowd that is always telling people what they can't do.

Re:What's insightful about this?

[Anonymous Coward]

You sound like a new age, crystal-gazing idiot. "Anything is possible, if you can dream it! Quit poisoning me with you negativity!"

No amount of novel innovation and discovery will ever make 2 + 2 equal 5. Not even with a really big value of 2.

  There are real, hard limits in our universe. Get used to it.

Re:What's insightful about this?

[theshowmecanuck]

nice troll... now tell me all the things we can't do.

This is an important point

[sean.peters]

I think it's a lot more likely that Moore's law (and technology expansion in general) is more likely to be following a logistic curve. It looks exponential for a while, but eventually levels out as you come upon fundamental limits to further growth.

As the kids say...

[sean.peters]

[citation needed]. Who were these physicists who allegedly said this?

For every complex problem...

[sean.peters]

... there's a simple (or if you prefer, simple simple) solution. And it's wrong. The issue is that many, many classes of computational problems are not amenable to breaking apart to solve in parallel. You pretty much have to do the steps in order. So parallelism only gets you so far.

Computational efficiency

[AlpineR]

Now, I just wish TFA actually gave a bit more information on the limit instead of just saying 10 quintillion times faster than todays computers.

Agreed. I'd like to see a performance factor that relates any real computer to the theoretical maximum.

I have toyed with defining such a factor myself. I was thinking of setting zero as the computational performance of ENIAC and 100 as the performance of the ultimate computer. Then any computer with performance between the two would be placed along a logarithmic scale. In addition to computational speed of the device, it would have to account of its mass/energy expenditure to keep the comparison balanced.

Then you could say "The new Mac Pro Dodecapod has a computational performance of 6.5". I wonder just how far we are along that scale. I think it would give new appreciation for how primitive our computer technology is compared to Nature's abilities.

(If somebody takes this idea and defines such a factor, you could do me the honor of calling it the Wagner factor.)

Oh NOES!

[Sinical]

Wow, who will provide my quantum-mechanically-accurate-universe-in-which-quantum-mechanically-accurate-porn-stars-work porn if we can have processors that are only 10,000,000,000,000,000 faster (per gram?) than currently (taking the 10^16 at face value). The children of the future will live impoverished lives of grievous destitution and horror!

"conduct yourself in a professional manor"

[Anonymous Coward]

I would venture a guess that you are all but un-employable, regardless of how intelligent you feel you are. If you are gainfully employed, I would appreciate it if you could conduct yourself in a professional manor, where you might be educated in the responsibilities of spelling by the nanny.

Re:Physicists said we could not exceed 2400 baud t

[whoisisis]

> Yet before we gave up on phone lines, the modem builders were giving us 56,000 baud connections.

Yeah, but isn't that done by compression of data? Send the compressed data with 28 kbaud.
Also if you look closely, the 56 kbaud is a theoretical upper limit, not generally met in reality.

On a phone line with only 4 kHz bandwidth, yeah, you can still only send about 28 kbaud.
Today, we use "phone lines" with much larger bandwidth.