I don't know but I've always found it odd to be called a "law", that gives too much scientific credit for something that was clearly derived by simple observation and guesswork and has no underlying science to prove it except that the statement held for some time.
In danger of being a bit too pedantic, there is no such thing as a 'real scientific law' either. Though I admit there is general understanding of what is implied by 'a scientific law', and I assume that's what you and the article is referring to.
I subscribe to Kurzweil's camp in that I think the halting of Moore's law will pressure a paradigm shift in computing, like the electromechanical, relay-based, vacuum tube, transistor paradigms, the integrated circuit paradigm will cease to be relevant when something like memristors or three-dimensional molecular computing comes into play.
> the integrated circuit paradigm will cease to be relevant when something like memristors or three-dimensional molecular computing comes into play.
What part of "integrated circuit paradigm" doesn't apply to memristors or three-dimensional molecular computing?
I ask because HP's Williams seems to say that memristors are firmly in the "integrated circuit paradigm", but maybe he means something different than you do.
Moore's law is very much rooted in photolithography which is a 2D process. And memristors do fit in this. So does the 3d computing that intel is doing now, actually, since it still has a flat product.
but if ever there were truly solid 3d computers then they might scale on a different curve, presumably sharper due to the cubic.
photolithography is already limited by heat density's. Going 3D sounds great but it's far less helpful than you might hope and clock speeds are dominated more by how long it takes a transistor to switch than how long it takes electricity to travel though those tiny wires.
It states a doubling time. (That page is about number of circuits, but there's another "law" about fab costs.) It doesn't specify a technology or geometry.
That doubling-time changed about 2005. Maybe 3D would change it, maybe it wouldn't.
Note that cubes of active elements are harder to cool than planes.
Moore's Law was about the amount of transistors you can put in a chip cheaply, so it may be coming to an end. But I don't think it will stop machines from getting faster - we are seeing lots of progress in multi-core usage even on desktop and in mobile computing (multicore smartphones), and using dedicated hardware outside the CPU socket/package (GPGPU) is getting normal too - those look like new ways of making faster personal machines.
People have been predicting the end of Moore's[1] law for years - since at least mid 90s.
Luckily today most people are at the point where they just don't need more power. Most people need enough to run a web browser and do a bit of word processing. They might need enough to open a spreadsheet or a presentation.
Most people would be happy with something like a good tablet and a monitor / keyboard dock.
It would be nice if the churn of "more power" could switch focus to "power efficient" or "better optimisations" or "better architecture" or even "future technology with advanced architectures".
Sure, there are people who need media streaming; or gaming; or compiling; or multicore crunching; or rendering farms; or etc. They'll always have machines. I guess parallelisation needs to improve a bit and clustering needs to get better.
[1] some of these people were using less formal definition of "double the 'power' / 'speed'" rather than more formal "double the number of transistors per IC".
I can't stand when I see people making this argument. What if people said that we had "enough" computing power back in 1999? Would you be happy with your Nokia candy-bar phone today? You'd never know because that's all you'd know!
We never have enough. The march of technology should be relentless. I for one can't wait to see how the guys at Intel/AMD/ARM or someone out of left field comes up with a way to get around this.
people have been saying that for years too. and they are always wrong. there are two things you are not considering: 1) that web browser or tablet still has to connect to a server that does the real work of the application and 2) the CPU needs of speech recognition, computer vision, and search are limitless. you seem to admit that when you say "they'll always have machines" but it's not "they" it's everyone. and FYI the chip industry has focused on efficiency since around 2000. the Atom line and ARM are not accidents.
There was enough power on the desktop for commercial-grade word processing and spreadsheets, at an affordable price, 30 years ago. Ignoring all the bells and whistles, there's actually not much you can do on a 2Ghz quad-core PC with 8G of RAM that you couldn't do on a 2Mhz 6502 machine with 32k or 64k of RAM.
The failure of the industry is, we have thousands of times more power - and we can't think of anything to do with it. And the keyboard of a Beeb or a C64 was much nicer than most modern keyboards too...
I don't think we have to worry too much about it until 2030 or so. At least until 2020 it should be a smooth ride, and then they'll probably stacks chips on each other or find some work-around for the next decade or so, until something else comes out.
In terms of comparing clock speed with, say, the brain: isn't it more the parallelization that we're lacking, rather than the speed? In other words lets say we could parallelize a computer network as complicated as the brain... with current processor speeds, couldn't we already re-create effective human intelligence?
I guess I mean to suggest it's the parallelization/inter-connectiviy and algorithmic challenge rather than the speed challenge that has so far avoided the singularity.
Something you're probably going to be interested in: Whole Brain Emulation. [0]
Human intelligence is IMO something that is referred to in pop culture as "emergent behavior" [1]. Problem is that you risk building the electronic equivalent of a cargo cult [2]. After all, on a cellular level of a nervous system, what's the difference between what happens in the brain of a chimp and that of a human? How detailed do you have to go to build a platform that can house intelligence?
On a philosophical level you kinda end up with a contradiction: is the human brain complex enough to understand/build the complexity of a human brain? Personally I'd be happier if research in this field was a bit more modest. I once read (somewhere in the 90s) that there was a researcher in the U.S. who was mapping every cell's behavior in the fly's "brain". With current tech, it would be feasible to implement that into a simulator and see what pops out. But doing the same for a human brain? Let's first try to come up with a definition for fuzzy meta-sensations such as intelligence and emotions that would translate to computer lingo.
I'd suggest it's not a contradiction in so far as... we're not talking about one human brain creating another. We're talking about the collaborative work of countless human brains adding on to each other's work. E.g. parallelization of brains is what enables great human strides, and so far seems quite limitless :)
We're talking about the collaborative work of countless human brains adding on to each other's work.
Every bit of progress in humanity is the work of an individual, either working individually or in group. Collaboration still revolves around individual contributions. In a lot of cases (e.g. group discussions and brainstorms) we get the impression that it's the group producing the result, but in reality it's incremental individual work.
You're claiming that collaboration can overcome the issue (complex brain understanding it's own complexity). I claim it can't : the ceiling of what can be achieved depends on what the smartest brain (so to speak) can comprehend and tie together.
An ant doesn't know how its colony functions, it takes a human to capture that. Same for us, it takes a more evolved brain to understand what makes us tick, we'll only sample aspects of it. A system can never be complex enough to understand its own complexity.
I don't think processor speed is the real problem when trying to recreate human intelligence. We don't even know the algorithm, otherwise we could simulate it at below-realtime speeds. And since we also don't know the exact structure of the brain in sufficient detail, brute-force simulation is also infeasible.
For a couple of decades we've been trading computational resources for programmer convenience. That worked very well when cores were rocketing up in speed year on year. But since we can't count on that, I think we'll be taking a harder look at our tools.
But I think we're going to have to take an even harder look at how we work. As processors stop getting faster, software won't become obsolete as quickly. We've all left a lot of mistakes behind in code killed in platform shifts. But now there may be no escape. I don't know what will come of that, but I'm sure interested to see.
2 years ago I worked at Intel and they talked about Moore's law and how people always said it would end, but it never does due to continued innovation. I think everyone knows they have a lot of people working on a lot of possible solutions to these physics problems. Intel is like a supercharged academia where hitting objectives matters. They don’t want engineers/physicists that say it cannot be done, those people are stuck on the outside writing posts like this one. Those posts come up continually, (do a search for “end of moore’s law 200X) and you will see articles written every year.
Here is a recent quote from an interview with Mark Bohr, Intel's senior fellow and director of process architecture and integration.
“"The end of Moore's law has always been 10 years away, and it will always be 10 years away," he said. He's been hearing predictions about the end of scaling since he joined the industry 30 years ago, so he isn't worried. He said 14nm is in full development and on track for manufacturing readiness in the second half of next year.”
For a while transistor density's doubled every 12 months. That stopped happening, then for a while they doubled every 18 months until that stopped happening. Now, if your willing to accept transistors doubling every 50,000 years as the same exponential growth as doubling every 12 months then sure it will continue for a long time. But, that's not what 'Moore's law' as originally stated.
Since 1975 it has been every 24 months. I have never heard the 12 month number until reading it on HN today.
Someone at Intel at some point said 18 months, but it was most likely a 1 off conversation/presentation rather than a restatement of the law and occured well after 1975.
The law is named after Intel co-founder Gordon E. Moore, who described the trend in his 1965 paper.[2][3][4] The paper noted that the number of components in integrated circuits had doubled every year from the invention of the integrated circuit in 1958 until 1965 and predicted that the trend would continue "for at least ten years".[5]
I wish people would stop citing wiki on HN, especially without reading the entire page. On the exact same wiki page under history it says:
Moore slightly altered the formulation of the law over time, in retrospect bolstering the perceived accuracy of his law.[17] Most notably, in 1975, Moore altered his projection to a doubling every two years.[18][19] Despite popular misconception, he is adamant that he did not predict a doubling "every 18 months." However, David House, an Intel colleague, had factored in the increasing performance of transistors to conclude that integrated circuits would double in performance every 18 months.[note 2]
Moore's law has been redefined several times already. At the beginning it was every 12 months, then 18, and now it's 24. What's being measured has also changed. It used to be frequency scaling that we measured, but that changed after the debacle of the Pentium 4.
But the essence of Moore's law has always remained: the exponential growth in the capability of ICs. The other relatively constant aspect of Moore's law is that people have always predicted it to last another 10 years. When Moore first formulated the law, that's approximately what he believed. So when somebody says that they expect Moore's law to end within 5-10 years, I take that with a big grain of salt.
What will kill Moore's law will be an unwillingness to continue to spend more money to build semiconductor plants. It used to cost only a few hundred thousand dollars to build an IC plant. This cost has increased exponentially to the point where a plant today costs $10 billion dollars. It's reasonable to suppose that a consortium of semiconductor manufacturers could invest $100 billion in a plant, but on current trajectories, a $1 trillion plant may be necessary to continue Moore's law in 10 years. Do you think that will happen?
The cost of the exotic silicon and post silicon alternatives are dramatically higher than todays costs, which are dramatically higher than yesterdays. Couple that with the fact that the majority of people have far more computing power than is necessary for their day-to-day existence (e.g. emailing, posting on FB and looking at cat pictures and/or pornography.) So costs continue to increase by some super-linear function but benefits follow a sigmoid curve.
The disconnect between marginal cost and marginal demand for computational power is what's going to knock Moore's Law off it's curve IMO, not some hard technical barrier. The futurists are always spending someone else's imaginary money, so they don't have to worry about stuff like that.
> The futurists are always spending someone else's imaginary money.
A $35 raspberry pi can outrun what used to be multi-million dollar mainframes. Are you sure that cost factors are relevant when the trend is so strongely toward lower costs? If prices drop, then lower utility uses become feasible.
Past returns do not indicate future performance. Technical innovation is slowing down, not speeding up. Raw, usable linear CPU performance at the fingertips of normal humans using personal computers has been essentially flat for half a decade, with the majority of material perf changes coming in the I/O subsystem, and yet still the CPU sits idle for the vast majority of its life.
I'm probably wrong, but it's still a reasonable position: we've seen technical stagnations before. I'm simply (and probably incorrectly) projecting the current trend I see, as an end user of technology: raw straigh-ahead cpu perf is stagnating and, unless I see real progress on AI (which I don't) I'm skeptical that a lot more of it will help most people doing most things, unless we are talking a lot lot more.
If some of the more exotic technologies get practical and cheap, I'm happy to be wrong, but I think it's important to note that Moore's Law was observational: he was describing what he saw happening rather than envisioning it. We no longer are seeing what he saw. A phase change may fix that, but that's very different than what we've been mining for the last half century, and much harder to project out.
Moore's Law in Silicon will eventually run out. This is true and is physically provable. And there are already a lot of smart minds (at Intel and at a few other companies and at many universities) developing post-silicon technologies. It is possible that we will one day shrink down to true molecular electronics using organic semiconductors or similar. As Feynman said, there's lots of room at the bottom. 3D molecular electronics are exciting for me---architectures will have to be redesigned from scratch, maybe RISCs too. It's an exciting time.
(I omit talking about quantum computing because that's currently mostly theoretical. But back in the day, Turing, Church, et al were talking about computability, etc when a real computer (von neumann architecture) did not exist yet. So maybe (I would be confident enough to say probably) down the line, we'll have a quantum computer, but it may not be soon.)
And it looks like memristors will hit the market around 2015, giving us much faster storage and nonvolatile RAM. According to HP, "We put the non-volatile memory right on top of the processor chip, and, because you’re not shipping data off-chip, that means we get the equivalent of 20 years of Moore’s Law performance improvement"
http://www.electronicsweekly.com/Articles/22/05/2012/53718/u...
Down the road a bit further, memristors could be used for neural-network coprocessors, since they function a lot like synapses.
The smooth progression of Moore's Law will probably get more jumpy, but the long-term trend goes back to mechanical adding machines.
I think it would take a lot longer than 5 years, even if node shrinking stopped dead.
You can always add more functionality to SoCs (eg better video decoding), tweak current cores (Intel gets more performance out of processor redesigns than node shrinks), and optimize your current node (power usage typically drops, and yield increases with time on the same node).
You would also get way more focus on GPGPU computing, with initiates like Intel's QuickSync. Lots of performance/watt to gain there, if there's no alternative.
Plus people making the hardware and software of phones and laptops need to sell units too, so they'll pick up some of the slack. Not every iPhone is a significant upgrade in terms of performance.
And lastly, a whole bunch of chips are being bought for reasons other than the old one is too slow; people are getting multiple computers, more gadgets, and in developing countries people are getting their first computers. Devices are becoming disposable. Laptops die and get replaced, even though even a Pentium M stacks up relatively well these days.
20+ years, though, you may have a point. I have no idea what that would look like. Though with depreciation on fabs slowing way down, even if sales were lower, the margins would look pretty damn good with today's prices. Plus each node shrink wouldn't be a potential disaster; the fab companies still around may even be more financially stable.
I'm not an electrical or thermal engineer, but even if we hit a wall as far as process sizes, couldn't they make bigger dies with more cores?
Like, if we can't get much smaller than a 12-nm process, couldn't we just make the die twice as big and put a big-ol heatsink on it to get to 128 cores or whatever?
It seems like architecture is still a bigger barrier than process size when it comes to getting >16 or >32 cores on the chip. How big will the current ring bus architecture that intel has scale?
costs for larger dies do not scale linearly. A chip with twice the area is at least 4x as expensive, if not more, depending on many factors like defect density and die packing. But you're right that this is the direction that the industry will move if process innovation stops. As a process matures, defect rates drop, making larger chips more feasible.
So basically, larger dies == greater chance a given chip has a defect, bigger cost and % of the wafer for each defective chip, and presumably more waste silicon on the wafer as well?
50 comments
[ 2.8 ms ] story [ 109 ms ] threadWhat part of "integrated circuit paradigm" doesn't apply to memristors or three-dimensional molecular computing?
I ask because HP's Williams seems to say that memristors are firmly in the "integrated circuit paradigm", but maybe he means something different than you do.
but if ever there were truly solid 3d computers then they might scale on a different curve, presumably sharper due to the cubic.
I don't see that "root".
http://en.wikipedia.org/wiki/Moores_law
It states a doubling time. (That page is about number of circuits, but there's another "law" about fab costs.) It doesn't specify a technology or geometry.
That doubling-time changed about 2005. Maybe 3D would change it, maybe it wouldn't.
Note that cubes of active elements are harder to cool than planes.
Then the article undermines itself by talking about 3d transistors and molecular valves and other possibly ways forward.
In the end, computational density is likely to be limited by heat transfer rates. Having your CPU melt itself is rarely desirable.
[0] http://news.cnet.com/2100-1006_3-6119618.html
(Disclaimer: don't know how the crowd here reacts to CNET articles)
http://en.wikipedia.org/wiki/Larrabee_(microarchitecture)
Luckily today most people are at the point where they just don't need more power. Most people need enough to run a web browser and do a bit of word processing. They might need enough to open a spreadsheet or a presentation.
Most people would be happy with something like a good tablet and a monitor / keyboard dock.
It would be nice if the churn of "more power" could switch focus to "power efficient" or "better optimisations" or "better architecture" or even "future technology with advanced architectures".
Sure, there are people who need media streaming; or gaming; or compiling; or multicore crunching; or rendering farms; or etc. They'll always have machines. I guess parallelisation needs to improve a bit and clustering needs to get better.
[1] some of these people were using less formal definition of "double the 'power' / 'speed'" rather than more formal "double the number of transistors per IC".
We never have enough. The march of technology should be relentless. I for one can't wait to see how the guys at Intel/AMD/ARM or someone out of left field comes up with a way to get around this.
The failure of the industry is, we have thousands of times more power - and we can't think of anything to do with it. And the keyboard of a Beeb or a C64 was much nicer than most modern keyboards too...
I guess I mean to suggest it's the parallelization/inter-connectiviy and algorithmic challenge rather than the speed challenge that has so far avoided the singularity.
Curious if others agree?
Human intelligence is IMO something that is referred to in pop culture as "emergent behavior" [1]. Problem is that you risk building the electronic equivalent of a cargo cult [2]. After all, on a cellular level of a nervous system, what's the difference between what happens in the brain of a chimp and that of a human? How detailed do you have to go to build a platform that can house intelligence?
On a philosophical level you kinda end up with a contradiction: is the human brain complex enough to understand/build the complexity of a human brain? Personally I'd be happier if research in this field was a bit more modest. I once read (somewhere in the 90s) that there was a researcher in the U.S. who was mapping every cell's behavior in the fly's "brain". With current tech, it would be feasible to implement that into a simulator and see what pops out. But doing the same for a human brain? Let's first try to come up with a definition for fuzzy meta-sensations such as intelligence and emotions that would translate to computer lingo.
[0] http://www.youtube.com/watch?v=kRB6Qzx9oXs
[1] http://en.wikipedia.org/wiki/Emergence
[2] http://en.wikipedia.org/wiki/Cargo_cult
I'd suggest it's not a contradiction in so far as... we're not talking about one human brain creating another. We're talking about the collaborative work of countless human brains adding on to each other's work. E.g. parallelization of brains is what enables great human strides, and so far seems quite limitless :)
We're talking about the collaborative work of countless human brains adding on to each other's work.
Every bit of progress in humanity is the work of an individual, either working individually or in group. Collaboration still revolves around individual contributions. In a lot of cases (e.g. group discussions and brainstorms) we get the impression that it's the group producing the result, but in reality it's incremental individual work.
You're claiming that collaboration can overcome the issue (complex brain understanding it's own complexity). I claim it can't : the ceiling of what can be achieved depends on what the smartest brain (so to speak) can comprehend and tie together.
An ant doesn't know how its colony functions, it takes a human to capture that. Same for us, it takes a more evolved brain to understand what makes us tick, we'll only sample aspects of it. A system can never be complex enough to understand its own complexity.
For a couple of decades we've been trading computational resources for programmer convenience. That worked very well when cores were rocketing up in speed year on year. But since we can't count on that, I think we'll be taking a harder look at our tools.
But I think we're going to have to take an even harder look at how we work. As processors stop getting faster, software won't become obsolete as quickly. We've all left a lot of mistakes behind in code killed in platform shifts. But now there may be no escape. I don't know what will come of that, but I'm sure interested to see.
Here is a recent quote from an interview with Mark Bohr, Intel's senior fellow and director of process architecture and integration.
“"The end of Moore's law has always been 10 years away, and it will always be 10 years away," he said. He's been hearing predictions about the end of scaling since he joined the industry 30 years ago, so he isn't worried. He said 14nm is in full development and on track for manufacturing readiness in the second half of next year.”
http://forwardthinking.pcmag.com/show-reports/297801-intel-t...
Someone at Intel at some point said 18 months, but it was most likely a 1 off conversation/presentation rather than a restatement of the law and occured well after 1975.
The law is named after Intel co-founder Gordon E. Moore, who described the trend in his 1965 paper.[2][3][4] The paper noted that the number of components in integrated circuits had doubled every year from the invention of the integrated circuit in 1958 until 1965 and predicted that the trend would continue "for at least ten years".[5]
http://en.wikipedia.org/wiki/Moores_law
Moore slightly altered the formulation of the law over time, in retrospect bolstering the perceived accuracy of his law.[17] Most notably, in 1975, Moore altered his projection to a doubling every two years.[18][19] Despite popular misconception, he is adamant that he did not predict a doubling "every 18 months." However, David House, an Intel colleague, had factored in the increasing performance of transistors to conclude that integrated circuits would double in performance every 18 months.[note 2]
But the essence of Moore's law has always remained: the exponential growth in the capability of ICs. The other relatively constant aspect of Moore's law is that people have always predicted it to last another 10 years. When Moore first formulated the law, that's approximately what he believed. So when somebody says that they expect Moore's law to end within 5-10 years, I take that with a big grain of salt.
What will kill Moore's law will be an unwillingness to continue to spend more money to build semiconductor plants. It used to cost only a few hundred thousand dollars to build an IC plant. This cost has increased exponentially to the point where a plant today costs $10 billion dollars. It's reasonable to suppose that a consortium of semiconductor manufacturers could invest $100 billion in a plant, but on current trajectories, a $1 trillion plant may be necessary to continue Moore's law in 10 years. Do you think that will happen?
The cost of the exotic silicon and post silicon alternatives are dramatically higher than todays costs, which are dramatically higher than yesterdays. Couple that with the fact that the majority of people have far more computing power than is necessary for their day-to-day existence (e.g. emailing, posting on FB and looking at cat pictures and/or pornography.) So costs continue to increase by some super-linear function but benefits follow a sigmoid curve.
The disconnect between marginal cost and marginal demand for computational power is what's going to knock Moore's Law off it's curve IMO, not some hard technical barrier. The futurists are always spending someone else's imaginary money, so they don't have to worry about stuff like that.
A $35 raspberry pi can outrun what used to be multi-million dollar mainframes. Are you sure that cost factors are relevant when the trend is so strongely toward lower costs? If prices drop, then lower utility uses become feasible.
I'm probably wrong, but it's still a reasonable position: we've seen technical stagnations before. I'm simply (and probably incorrectly) projecting the current trend I see, as an end user of technology: raw straigh-ahead cpu perf is stagnating and, unless I see real progress on AI (which I don't) I'm skeptical that a lot more of it will help most people doing most things, unless we are talking a lot lot more.
If some of the more exotic technologies get practical and cheap, I'm happy to be wrong, but I think it's important to note that Moore's Law was observational: he was describing what he saw happening rather than envisioning it. We no longer are seeing what he saw. A phase change may fix that, but that's very different than what we've been mining for the last half century, and much harder to project out.
From the very beginning, it has been economically feasible component density, and as far as I know, that has not changed. Look at this graph for example: http://en.wikipedia.org/wiki/File:Transistor_Count_and_Moore...
(I omit talking about quantum computing because that's currently mostly theoretical. But back in the day, Turing, Church, et al were talking about computability, etc when a real computer (von neumann architecture) did not exist yet. So maybe (I would be confident enough to say probably) down the line, we'll have a quantum computer, but it may not be soon.)
And it looks like memristors will hit the market around 2015, giving us much faster storage and nonvolatile RAM. According to HP, "We put the non-volatile memory right on top of the processor chip, and, because you’re not shipping data off-chip, that means we get the equivalent of 20 years of Moore’s Law performance improvement" http://www.electronicsweekly.com/Articles/22/05/2012/53718/u...
Down the road a bit further, memristors could be used for neural-network coprocessors, since they function a lot like synapses.
The smooth progression of Moore's Law will probably get more jumpy, but the long-term trend goes back to mechanical adding machines.
You can always add more functionality to SoCs (eg better video decoding), tweak current cores (Intel gets more performance out of processor redesigns than node shrinks), and optimize your current node (power usage typically drops, and yield increases with time on the same node).
You would also get way more focus on GPGPU computing, with initiates like Intel's QuickSync. Lots of performance/watt to gain there, if there's no alternative.
Plus people making the hardware and software of phones and laptops need to sell units too, so they'll pick up some of the slack. Not every iPhone is a significant upgrade in terms of performance.
And lastly, a whole bunch of chips are being bought for reasons other than the old one is too slow; people are getting multiple computers, more gadgets, and in developing countries people are getting their first computers. Devices are becoming disposable. Laptops die and get replaced, even though even a Pentium M stacks up relatively well these days.
20+ years, though, you may have a point. I have no idea what that would look like. Though with depreciation on fabs slowing way down, even if sales were lower, the margins would look pretty damn good with today's prices. Plus each node shrink wouldn't be a potential disaster; the fab companies still around may even be more financially stable.
Like, if we can't get much smaller than a 12-nm process, couldn't we just make the die twice as big and put a big-ol heatsink on it to get to 128 cores or whatever?
It seems like architecture is still a bigger barrier than process size when it comes to getting >16 or >32 cores on the chip. How big will the current ring bus architecture that intel has scale?
So basically, larger dies == greater chance a given chip has a defect, bigger cost and % of the wafer for each defective chip, and presumably more waste silicon on the wafer as well?