Slashdot Mirror
Intel Researchers See Moore's Law Becoming Obsolete
prostoalex writes "A paper, published by Intel researchers, claims we might be the witnesses of Moore's Law becoming obsolete, as the rate of shrinkage for transistors goes lower with each year. In 2018 we might be able to get the chips manufactured with 16-nanometer technology, then one or two more manufacturing processes will shrink it even further, but after that we're facing the physical limits."
Silicon is, indeed, close to its limit, but that does not mean semiconductors are.
This Wired article, which I'm sure many of you have read, details how new industrially-produced diamonds, thanks to their cheap price and purity (most importantly, being absolutely identical to each other), along with research done by both the government, several corporations, and possibly Intel, may make unbelievably fast systems powered by diamond semiconductors possible.
Some interesting quotes:
Also, a rather ironic one from Intel themselves:
Silicon is dead. Long live diamonds!
Engineers will be able to continue the shrink for another 15 years based on what we know now. However, the cost for designing setting up manufacturing for a chip will continue to increase exponentially. It will only be worth the money to do this for a part that can be sold in the billions, and there will be few such parts. The end will come not because the technologists can't reduce feature sizes any further, but because no one will be willing to sink an investment equal to the GDP of a mid-sized country into a fab.
At least, that's the case for CMOS silicon chips. To get Moore's Law to continue to operate in a meaningful way, something completely new is likely to be needed: maybe molecular gates that self-assemble or something equally exotic.
Even if there were no way to manufacture chips smaller/faster than the ones we have today, there are always going to be refinements in the manufacturing process, making chips cheaper and cheaper. There are always supercomputers. Perhaps, also, we could find a way to really minimize waste heat, allowing many CPUs per board.
It's also possible that DNA computation and other kinds of biocomputing are going to come along. These have the advantage of being gigantically parallel; they would possibly be good for tasks that are not latency sensitive but require immense brute force.
I'm satisfied that we have enough axes of advance to keep progress moving forward. Remember, computers have only been around for a very short while; I refuse to believe that we hit on the fitness maximum on the first try; there have to be technologies out there that are far faster/cooler/smaller.
Intolerance for ambiguity is the mark of the authoritarian personality.
I remember sitting in a lecture in 1997, where some luminary from IBM predicted the death of Moore's Law in 10 years. Now it's 2003 and the death of Moore's Law is being predicted in 15 years.
Technologically, there will probably be enough clever ideas to take chip manufacturing beyond the point where it is no longer economical to make such fast processors. Consider that in 1980, a handful of engineers could sit down with pencil and paper and design a microprocessor. Today it takes teams of PhDs in physics, math, and engineering to do the same, in multi-billion-dollar facilities with the latest design tools and techniques. One day the buying public will realise that e-mail and word processing does not need a bazillion gigahertz, and gamers will have photorealistic animation with excellent AI. The chip maker will not make back the investment on a fab plant, and on that day Moore's Law will be dead, not for physical reasons but for economical ones.
Toronto-area transit rider? Rate your ride.
The number of papers publicly published proclaiming the "real soon now" end of Moore's law will double every 18 months.
Irritable, left-wing and possibly humorous bumper stickers and t-shirts
This is how you visualize an electron tunnelling across a gate:
Heisenberg's uncertainty principle says that we can't know an electron's position accurately. There's always a little bit of uncertainty about where it is. So, imagine the position of an electron not as a point, but as a little 'O'. That circle is the area that the electron could be. At any time it could be in any random place in that circle.
Now, if the 'O' is centered on the edge of one side of the gap, and the gap is bigger than the circle radius, then the electron has zero probability of crossing the gap. But, once the gap is smaller than the radius of the circle, then you've got parts of both sides of the gate within the area of the circle. Since the electron can appear randomly anywhere inside the circle, that means that sometime that electron will appear on the other side of the gate. As the gates get smaller, the probability that the electron will randomly appear on the other side of the gate goes up, until so many electrons are crossing the gate that we can't tell if the thing is on or off.
This is America, damnit. Speak Spanish!
4 and 10 Ghz is a huge jump. I doubt Intel would release them that close together. It would be horrible marketing sense. Why make such a big bang jump to 4 and 10 when Intel can suck much more money producing a 4 Ghz then a 5 Ghz and the 6 and so on. Indeed I am questioning your source, but time will tell if you are correct about these releases. As far as Moore's law: In the past when people have said Moore's law must stop it was because researchers were having harder and harder times finding ways to product smaller chips. Now we are getting close to the point that we are arranging the silicon semiconductors atom by atom. Once your organizing atoms you physically cannot do much more. You cannot work with smaller components than on an atom by atom basis. Researchers have trouble even isolating the constituent parts of an atom, and the components of an atom are still highly theoretical. And those components that have been identified are highly unstable. Supposedly though there is something called quantum computing. I don't understand it but maybe quantum computing which doesn't use transistors (as far as I know) will be the future.
From Intel's website: "Moore observed an exponential growth in the number of transistors per integrated circuit and predicted that this trend would continue. "
To be a tiny bit pedantic, Moore's original paper talked about the number of transistors per integrated circuit at any given price point. You can always stick more transistors on the chip if you're willing to throw sufficient amounts of money at the problem, but to get those transistors for a reasonable price is another matter.
FWIW, Moore's original hypothesis was that the transistors/$ would double every 12 months, so his "law" hasn't been correct for quite some time. We had been seeing a doubling of transistors about every 18 months for a while, but now it's more like every 24 months. With the current troubles that Intel, AMD and IBM all seem to be having at implementing their new 90nm manufacturing process, it seems likely that the pace will continue to slow.