If you're looking for pundits with an end date for Moore's Law, you don't have to look far. You also don't have to look far to find a gaggle of loonies who just knew the world was ending in Y2K, so make of that what you will. The latest duo looking to call the demise of the processor mantra that has held true for two score comes from Boston University, with physicists Lev Levitin and Tommaso Toffoli asserting that a quantum limit would be achieved in around 75 to 80 years. Scott Aaronson, an attention-getter at MIT, expects that very same limit to be hit in just 20 years. Of course, there's plenty of technobabble to explain the what's and how's behind all this, but considering that the brainiacs of the world can't even agree with Gordon Moore's own doomsday date, we're choosing to plug our ears and keep on believin' for now. Bonus video after the break.
Moore (co-founder of intel) was just describing his observations about the microprocessor industry at the time (1965). He never intended his prediction to be taken as law.
It acts as a product life-cycle goal and not the defacto rule of performance. If a company like intel or IBM expects to remain viable then that company would have to find a way to double the performance of it's products every 18 months. Otherwise a competitors product would likely over take them until. For that reason it's more of an self-fulfilling prophecy then a law.
We don't have people on the moon, and I'm not referring to the 'faked landing' but the current facts, we can't put people on the moon in any reasonable amount of time currently. And I think the timeframe mentioned for the end of moore's law takes all possible advancements into account, coupled with the simple laws of nature, and doesn't seem unreasonable
75 to 80 years? By that time we'll come up with some other new theories that will show how to get it even smaller. Quantum theory, itself, was barely in its infancy 80 years ago.
@cocoviper: That's assuming we're still using molecules in 80 years. That's a LONG way off(longer than most people here will be alive), so making any assumptions about what the world will be like or what tech we'll be using is foolish.
@duggy: A: that's not true, electrons, protons and (obviously) neutrons are all matter as well. B: why are you limiting things to matter?
Like I said before, 75 to 80 years is a REALLY long time. Long enough, in fact, that you're going to have to take into account massive amounts of scientific theory that nobody has even thought of yet.
@tacomanmcjhonny: Right, I keep forgetting that to people like you, using logic makes one a douche. Then again, you also apparently think that because electrons and protons carry a charge, they don't have mass. So I guess nobody should really pay attention to what you think.
You'd have to be fairly optimistic to believe that in 80 years we would have processors made from pure electromagnetic energy that would interact with other electromagnetic and electric energy to perform logical calculations.
@duggy: I dunno, 80 years ago it was 1929 and they probably would have called someone insane if they said that we'd have the internet, that we would have people on the moon, that we accelerate particles to near the speed of light, that we regularly convert matter into energy, that we could use superconductors to create frictionless coils, etc. My point with this is that you can't use physics to predict where new discoveries in physics will be in 80 years. You could say given our current knowledge, this is the peak, but claiming that we won't have some new knowledge that unlocks even faster speeds/more transistors than is currently thought to be physically possible by then is ridiculous.
The molecular barrier is not the end of Moore's Law. You are making the assumption that shrinking the electronic components of microprocessors is the only way to increase performance.
Processor speeds haven't really gone up. They just add more cores and up the front side bus.
The reality is that most computer users don't need much more speed. They need applications that can make good use of the existing speed and the multiple core architecture.
The fact is, the more speed we get, the more things that can, and will eventually be done... and then even those users that you claim don't need more speed will find themselves using that extra speed because of new applications that just weren't possible before.
i disagree with these two physicist (though who the hell am i to disagree with them) though the end of moores law maybe coming to an end for silicon based processors it will not for other types of processors that will be made in the future. for example scientist are already trying to make processors out of graphene which if successful will extend moores law for a very long time since you can make graphene chips more compact and not have to worry about heat. also there's quantum computing and who knows what else.
They were talking about the absolute minimum time it could possibly take a computer to perform an operation:
"Levitin and Toffoli present an equation for the minimum sliver of time it takes for [the most basic task a quantum computer could carry out] to occur. This establishes the speed limit for all possible computers."
That's just 0 --> 1 inside a quantum computer. There's no limit on how many of those operations can be carried out in parallel, thus computers are free to become self-aware and murder us all.
Well lets see, if that's 20 years, doubling speed every 18 months. That's computers that are about 4000 times faster than they are now. Yeah, we're screwed.
Well, there's kinda two moore's law's... there's the literal one (about how fast density of transistors on chips goes up), and then there's the one about how fast the power of computing goes up. They're related, but most people really mean the latter when they talk about moore's law ... even though only the former is correct. And, in the greater scope of things, the common/less-correct definition is the one that really matters. That's the one that's about how our society evolves as our technology evolves. The other one (the more correct one) is merely about one particular form of technology (a form of technology that we'll hopefully out-grow ... though, if petroleum based fuels are any indication, we may hang on to it for a lot longer than is necessary/healthy).
Anyway... my point is: if you go with the more common (and less correct) version of moore's law ... yes, at some point we'll reach a point where semi-conductor transistors wont be able to shrink any further, and semi-conductor chip density will stop increasing. At that point, the strict version of moore's law is over ... but the common version of moore's law is not necessarily over.
If we find a new basis for computing (optical logic based computing, holographic computing, quantum computing, magnetic computing, etc.), then we may find that, with a slight adjustment (jink up or jink down and/or increase or decrease in the exact steepness of the curve) we'll still be on an exponentially increasing curve in computing power.
That's bull. You know how hard it is to glue all those tiny transistors onto a chip? For that matter, do you realize how hard it is to open a bag of new transistors without sneezing and blowing them all over your work space?
Moores law only requires you to double transistors, you don't HAVE to shrink the thing at the same time. They can start building upward in 3-D, maybe future processors will be shaped more like a cube.
What is it about Moore's law that really makes it 'hold'? I think its actually just the result of intel's R&D investment strategy...
If intel wanted to triple performance in 18 months, they could, just spend more money on the problem. But, they've found that 2x/18mo is good enough.
And as a response to the individual above who said that 'most users dont need more speed', I'll agree with that - for home users - however, go ask a national lab how much they need to do what they want and they'll tell you 1000x more than we have now.
It's obvious that Moore's Law will end along with the rest of the world in 2012... Do people really have nothing better to do than to try and predict the future? Now if you'll excuse me I have to watch television so I know what the weather will be like tomorrow.
i'm tellin you guys...when computers will become self aware, they won't kill us all or take over the earth, they'll simply party all night long and cry at journey's don't stop believing song:((...i know, it gets me to
These scientists are just blowing up smoke up our asses, two hundred years ago we weren't even aware of the basic structure of atoms let alone aware of quantum physics. Yes their equation might hold true in terms of the math and physics we posses today, however who is to say that we won't find even more ingenious ways of computing information.
They are also specifically referring processor die shrinkage, we live 3 dimensionally. If worst comes to worst go wider, longer, taller, heck we might go back to the good old days of computers being the size of houses one day.
One thing is for sure technology will ALWAYS continue to advance, beyond our wildest imagination.
"Moore's Law describes a long-term trend in the history of computing hardware, in which the number of transistors that can be placed inexpensively on an integrated circuit has doubled approximately every two years"
Doesn't matter. By 2050 we'll reach the technological singularity and robots will be taking over. Not sure they'll care about moore's law sense they'll be 1000x smarter than us. they will probably laugh at our tiny intellects. Just hope they don't have a taste for human blood.
A limit can be reached, only when assuming that space is not warped to allow for electricity to effectively travel faster than the speed of light. When this is accomplished, computers will literally be able to finish things before the user even begins to interact with them. After this becomes commonplace, it will become apparent that intention, action, past, and present, are only related by the perception of them. This will then be forgotten, and we will end up going in reverse/forwards/whatever-"direction"-wards, into infinity.
I have been in this situation so many times, it's not even funny, anymore.
Why is this so loony? Processor speed hasn't moved much past the 3,06 GH since 2002! And until developers solve the current thermal limitations inherent in CPU's, they won't.
Now that we've thrown 'em off the trail, use the form below to get in touch with the people at Engadget. Please fill in all of the required fields because they're required.
posted Oct 20th 2009 6:01PM
Hold on to the feeling!!!!
FASTER SMALLER MORE POWER!!!!!
That's what she said?
Hes a cool guy and all... except one thing... he kind of beats puppies with kittens...
It will stay alive just in a diamond platform the silicon platform will die but diamonds are the key to our future processing needs :)
Moore (co-founder of intel) was just describing his observations about the microprocessor industry at the time (1965). He never intended his prediction to be taken as law.
It acts as a product life-cycle goal and not the defacto rule of performance. If a company like intel or IBM expects to remain viable then that company would have to find a way to double the performance of it's products every 18 months. Otherwise a competitors product would likely over take them until. For that reason it's more of an self-fulfilling prophecy then a law.
No singularity ? Damn.
Shun the nonbelievers. Shunnnnnnnn.
you go on and on and on.... moore's law?
1st at what?? gaggle of loonies?
"gaggle of loonies who just knew the world was ending in Y2K"
Hey! I resemble that remark!
We don't have people on the moon, and I'm not referring to the 'faked landing' but the current facts, we can't put people on the moon in any reasonable amount of time currently.
And I think the timeframe mentioned for the end of moore's law takes all possible advancements into account, coupled with the simple laws of nature, and doesn't seem unreasonable
to be drowned... O I sincerely hope so.
75 to 80 years? By that time we'll come up with some other new theories that will show how to get it even smaller. Quantum theory, itself, was barely in its infancy 80 years ago.
I believe the 75 to 80 years assumes we hit the molecular limit. I.e. you can't make a physical processor smaller than the molecules it's made up of.
@cocoviper: That's assuming we're still using molecules in 80 years. That's a LONG way off(longer than most people here will be alive), so making any assumptions about what the world will be like or what tech we'll be using is foolish.
The closer we get to the subatomic level, the closer we get to the Higgs Boson traveling through time to kill us all. Moore's law will kill us all.
Huh, looks like the down-rank bot is back. It went on hiatus for most of today, so welcome back!
Considering that all matter in the universe is made of atoms, what else are we going to be using?
@duggy:
A: that's not true, electrons, protons and (obviously) neutrons are all matter as well.
B: why are you limiting things to matter?
Like I said before, 75 to 80 years is a REALLY long time. Long enough, in fact, that you're going to have to take into account massive amounts of scientific theory that nobody has even thought of yet.
duggy:
Photons
@ duggy
Cardassians
Um, why are all of Mark's posts downranked?
Is it the "in" thing to do?
We dont like mark... hes a douche... and p+ and e- are charges
Its p and n. p for holes, and n for free electrons ^_^
@tacomanmcjhonny: Right, I keep forgetting that to people like you, using logic makes one a douche. Then again, you also apparently think that because electrons and protons carry a charge, they don't have mass. So I guess nobody should really pay attention to what you think.
@cocoviper - Yeah - nature is only "SO" efficient.
You'd have to be fairly optimistic to believe that in 80 years we would have processors made from pure electromagnetic energy that would interact with other electromagnetic and electric energy to perform logical calculations.
@duggy: I dunno, 80 years ago it was 1929 and they probably would have called someone insane if they said that we'd have the internet, that we would have people on the moon, that we accelerate particles to near the speed of light, that we regularly convert matter into energy, that we could use superconductors to create frictionless coils, etc. My point with this is that you can't use physics to predict where new discoveries in physics will be in 80 years. You could say given our current knowledge, this is the peak, but claiming that we won't have some new knowledge that unlocks even faster speeds/more transistors than is currently thought to be physically possible by then is ridiculous.
The molecular barrier is not the end of Moore's Law. You are making the assumption that shrinking the electronic components of microprocessors is the only way to increase performance.
What about innovations like: multiple cores, on-chip memory controllers, branch prediction, pipeline efficiency, order-of-operation procedures, neural network engines, distributed computation, quantum computing, etc..?
Don't underestimate human ingenuity
Too busy listening to Journey to make a witty comment on your fail.
well,
Processor speeds haven't really gone up. They just add more cores and up the front side bus.
The reality is that most computer users don't need much more speed. They need applications that can make good use of the existing speed and the multiple core architecture.
Processor speed != Processing power.
That's a pretty dumb comment.
The fact is, the more speed we get, the more things that can, and will eventually be done... and then even those users that you claim don't need more speed will find themselves using that extra speed because of new applications that just weren't possible before.
i disagree with these two physicist (though who the hell am i to disagree with them) though the end of moores law maybe coming to an end for silicon based processors it will not for other types of processors that will be made in the future. for example scientist are already trying to make processors out of graphene which if successful will extend moores law for a very long time since you can make graphene chips more compact and not have to worry about heat. also there's quantum computing and who knows what else.
They were talking about the absolute minimum time it could possibly take a computer to perform an operation:
"Levitin and Toffoli present an equation for the minimum sliver of time it takes for [the most basic task a quantum computer could carry out]
to occur. This establishes the speed limit for all possible computers."
That's just 0 --> 1 inside a quantum computer. There's no limit on how many of those operations can be carried out in parallel, thus computers are free to become self-aware and murder us all.
Unless we go quantum, the idea of electrons accidentally going through that 11nm wall makes it seem like there has to be a limit in our future.
(Can quantum go any faster than quantum or are their levels of speed in quantum computing?)
Yeah and we'll reach 11nm in about 5 years so I hope they have some solution.
Apparently you weren't
moore's law has BEEN dead. 2009 and we STILL aren't in flying in fuckin laser shooting suburbans?! gimme a break -___-'
Well lets see, if that's 20 years, doubling speed every 18 months. That's computers that are about 4000 times faster than they are now. Yeah, we're screwed.
http://en.wikipedia.org/wiki/Moores_Law
Read up on this link, Moore's Law has nothing to do cars, flying or lasers.
Well, there's kinda two moore's law's... there's the literal one (about how fast density of transistors on chips goes up), and then there's the one about how fast the power of computing goes up. They're related, but most people really mean the latter when they talk about moore's law ... even though only the former is correct. And, in the greater scope of things, the common/less-correct definition is the one that really matters. That's the one that's about how our society evolves as our technology evolves. The other one (the more correct one) is merely about one particular form of technology (a form of technology that we'll hopefully out-grow ... though, if petroleum based fuels are any indication, we may hang on to it for a lot longer than is necessary/healthy).
Anyway... my point is: if you go with the more common (and less correct) version of moore's law ... yes, at some point we'll reach a point where semi-conductor transistors wont be able to shrink any further, and semi-conductor chip density will stop increasing. At that point, the strict version of moore's law is over ... but the common version of moore's law is not necessarily over.
If we find a new basis for computing (optical logic based computing, holographic computing, quantum computing, magnetic computing, etc.), then we may find that, with a slight adjustment (jink up or jink down and/or increase or decrease in the exact steepness of the curve) we'll still be on an exponentially increasing curve in computing power.
Moore's law was broken ever since Intel started developing their stuff by torturing aliens
The i7 was actually developed 5 years ago. They are only releasing it now because there wasn't any competition to Intel chips during the time
I like this theory. Very much.
That's bull. You know how hard it is to glue all those tiny transistors onto a chip? For that matter, do you realize how hard it is to open a bag of new transistors without sneezing and blowing them all over your work space?
Moores law only requires you to double transistors, you don't HAVE to shrink the thing at the same time. They can start building upward in 3-D, maybe future processors will be shaped more like a cube.
3D processors... 3D processors... Ummm, why does it remind me of: http://www.uncarvedblog.com/files/page1_blog_entry4_2.png :D
The world did end on Y2K but we were transferred to a parallel world by a secret group sworn to defend us. :)
Every time someone says 1st, a basket of kittens dies.
I hope you understand the consequences of your actions.
What is it about Moore's law that really makes it 'hold'? I think its actually just the result of intel's R&D investment strategy...
If intel wanted to triple performance in 18 months, they could, just spend more money on the problem. But, they've found that 2x/18mo is good enough.
And as a response to the individual above who said that 'most users dont need more speed', I'll agree with that - for home users - however, go ask a national lab how much they need to do what they want and they'll tell you 1000x more than we have now.
The dude looks like a chick in the leopard-print shirt.
That's Steve Perry and he slayed that leopard with his golden voice.
где гена?
The way technology is developing, we'll soon be able to predict the end of technology!
It's obvious that Moore's Law will end along with the rest of the world in 2012...
Do people really have nothing better to do than to try and predict the future? Now if you'll excuse me I have to watch television so I know what the weather will be like tomorrow.
It's not a law. It's more of a guideline.
i'm tellin you guys...when computers will become self aware, they won't kill us all or take over the earth, they'll simply party all night long and cry at journey's don't stop believing song:((...i know, it gets me to
Bah... Moore's Law isn't ending until we lose the need to have more power.
We will always discover new materials... and if they don't already exist, we'll make them.
There's a helluvalot of money in chips and a helluvalot of scientists and engineers in the world trying to out-do each other.
These scientists are just blowing up smoke up our asses, two hundred years ago we weren't even aware of the basic structure of atoms let alone aware of quantum physics. Yes their equation might hold true in terms of the math and physics we posses today, however who is to say that we won't find even more ingenious ways of computing information.
They are also specifically referring processor die shrinkage, we live 3 dimensionally. If worst comes to worst go wider, longer, taller, heck we might go back to the good old days of computers being the size of houses one day.
One thing is for sure technology will ALWAYS continue to advance, beyond our wildest imagination.
"Moore's Law describes a long-term trend in the history of computing hardware, in which the number of transistors that can be placed inexpensively on an integrated circuit has doubled approximately every two years"
Way too many syllables in these jokers' names to ever catch on...
Yeah, but what they lack in shortness they make up for in recycling the same letters. It evens out.
Haha, good point. Who knew male pornstars could also have a knack for science?
Doesn't matter. By 2050 we'll reach the technological singularity and robots will be taking over. Not sure they'll care about moore's law sense they'll be 1000x smarter than us. they will probably laugh at our tiny intellects. Just hope they don't have a taste for human blood.
Yeah but they will never know what it means to love and that will ultimately be there downfall.
Oh, and if computers are so smart why can't they tell me that I should have said "their" and not "there"?
Being logical they\d do away with politicians and the religious, not something that causes a great deal of anxiety with me.
Every time someone takes umbrage with somene who says "first" I kick a puppy.
A limit can be reached, only when assuming that space is not warped to allow for electricity to effectively travel faster than the speed of light. When this is accomplished, computers will literally be able to finish things before the user even begins to interact with them. After this becomes commonplace, it will become apparent that intention, action, past, and present, are only related by the perception of them. This will then be forgotten, and we will end up going in reverse/forwards/whatever-"direction"-wards, into infinity.
I have been in this situation so many times, it's not even funny, anymore.
Why is this so loony? Processor speed hasn't moved much past the 3,06 GH since 2002! And until developers solve the current thermal limitations inherent in CPU's, they won't.
Hey Marc - you are really not comparing a Pentium II - 3Ghz processor with the 3Ghz from a single Core of a current Multi-Core Processor are you?
That's like comparing the V8 engine in a Corvette from 1953 to a V8 engine in a 2009 Corvette. Doesn't work that way.
Didn't you hear that the Mhz/ Ghz stuff from 5-10 years ago was marketing - bad marketing at that?
Multi core is a hack thought up exactly BECAUSE they reached a limit, and don't use car (or sports) analogies on engadget please.
@OneLove - and then 9/11 happened - kills your theory.
All good things come to an end.