Slashdot Mirror
0.01 Micron Process?
hypo writes "According to a recent ZDNet article, IBM is developing a technique called "V-Groove", that allows the channel lengths of transistors on chips to be 10 nanometers (0.01 micron) and below. Currently, most companies use a 0.18 micron or 180 nanometer process. This is certainly a giant leap. The only caveat is that IBM is not planning to use this in large chips (i.e., processors) for 10 to 15 years. However, this is still quite revolutionary because most people thought that a 0.02 process would be the fundamental minimum. This all shows that Moore's law can perhaps hold true in the future. This article also discusses Carbon Nanotubes, which might research market faster than experts had previously thought."
Someone please correct me if I'm wrong (I certainly might be.. I'm not intimately familar with microelectronics engineering), but I thought what we currently associate with chip die processes are the trace widths, not the channel length.
Trace width is the width of the conductors connecting different transisitors on the chip. This is important because a smaller trace width means that the whole chip is scaled down, including the spaces between the traces. This raises capacitance between parallel wires and causes the posibility of cross-talk.
As for channel length, the article says:
Channel length represents the distance electricity needs to travel through a transistor, shorter transistors lessen the distance traveled, delivering greater performance.
While this is related to performance (specificly, switching timings), I am not sure if it is related to trace width at all. The ZDNet article may be mistakenly associating the two.
Also, I think that one may be able to vary the trace width and the channel lengths independently. If that is the case, we may have performance increases from channel lengths even if we hit a wall when it comes to trace widths.
Can someone with some microelectronic background clarify these issues?
Thanks.
Using your sig line to advertise for friends is lame.
Encoding at under 20x? try to find the GoGo encoder webpage. Assuming you have a faster processor (I have an Athlon 700 running WinNT and I can encode at 30x), you should be able to encode quickly and nicely. The Win32 counterpart encodes at 160Kbps and pretty much rocks da haus compared to all my other encoders.
You are right regarding the plasma part.
However, the process I described is also called bremsstrahlung. I used to work in the space industry (I am now doing astrophysics), and one of the reasons why thick Aluminium is not used as radiation shielding in space is because of bremsstrahlung effects from high energy protons.
Mode (3) smart-aleck mode. Press * to return to main menu.
What if you turn off optimisations on your C++ compiler? I know that VC++ does a good ammount of stuff toward the end of speeding up its output (it optimizes much better than C++Builder, for instance)
:)
Doesn't matter. Object Pascal compiles 10-100 times faster than Visual C++ with all optimizations turned off. The speed comes from a few places:
1. C++ programs tend to be idiotic with the include files. A 10,000 program may include 500,000 lines of includes. Object Pascal has a much nicer module system.
2. Object Pascal has a much cleaner syntax than C++ and doesn't need a preprocessing step.
3. The Object Pascal compiler is a very nice piece of programming
this technology would more likely be used for cell phones, etc, not desktop computers if theyre just now making a transistor on this scale, i would guess more than 10-15 years. most large companies are using 5 year old technology, and this is something that would require a lot more changes than redesigning stuff from 5v to 1.8v
handhelds, cellphones, battery life could be greatly extended using lvds and this type of size reduction
There's also a marketing issue -- As a company, you want to keep one step ahead of the competition. You also want to get the biggest bang for your research buck. If you get too far ahead of the competition, you won't be able to use, and make money off of, some of your other research. It's also nice to have an 'ace in the hole' for when they threaten to overtake you in another area.
Finally there's the simple lead time for going from producing a .01Micron straight line to producing a 100-million transister CPU from said technology -- and doing it in good quantity with high reliability.
-----
That having been said, I remember a story from a Nortern Telecom tech about the (relatively) early days of optical fiber. One of the labs claimed to have produced a really high-caliber optical repeater laser (about the size of a large grain of sugar). The production of the units was fobbed off on a Japanese company because the company big-wigs didn't believe lab staff that it could be done well using local resources.
Well the Japanese company messed up the order, (they weren't sensitive enough -- a prime specification) and the Exec turned to the lab and essentially said 'we need that order NOW -- Please do it with the lab equipment (no time to build a fab facility at this point).
Well, the lab made such high quality units that they were TOO sensitive. They were reacting to noise from the other electronics (which wasn't expecting such high quality in the repeater laser). Rather than re-design the electronics they went back to the lab and asked them to purposefully crank down the sensitivity of the lasRs.
Moral of the story: If IBM really HAD to get that stuff out the door in 18 months they could proabably do so. Chances are, however, that they can't see the long-term financial benefit of doing so.
Free Software: Like love, it grows best when given away.
And if you keep halving size, you get there in log time, ie sooner than you'd think.
My completly uninformed opinion is that silicon has a couple of decades left at best, but computation in general has a century or so before it runs up against the minimum scales and maximum efficiencies of matter and light.
My Karma: ran over your Dogma
StrawberryFrog
Well, actually, there will also be 0.15 um, which is what AMD is going to do with the next Athlon shrink. They claim they will be able to do it faster than 0.13 um, so they will be quicker to market. Who really knows ...
The next reduction in size will be to .13 micron. Intel is planning to make this transition on the P-III and P-4 chips in about Q3 of 2001. Slashdot posted this about the coming chips and micron size reductions. CNET has a story which is what the slashdot story is about. The CNET story though comes from this story of InQuest Market Research. Hope you like chip road maps as much as I do :)
My bad, you're completely right....I was thinking of virtual 86 mode.
Email me.
Don't trust anyone over 90000.
+++ATH0
yeah, but i figure that's ok, since the government's run by aliens anyway
now if they could just point that big dish at arecibo at washington, maybe they'd start getting some results...
Just because the technology is available doesn't mean it's *readily* available. Look at .18-micron chips. IBM is the only company that can make them on a somewhat consistent basis. They have about a 90% yield when it comes to making them. All other companies have about a 50% yield, if that. So just because a company can produce one working chip, that doesn't mean it will be able to efficiently produce enough to start selling. It could take 10 years for them to come up with a better process of producing them.
Actually, Moore as not talking about speed or copmutational power when he said "doubling," he was actually talking about the number of transistors on a chip. And that can't go on forever because transistors can only get so small and no one wants a four square meter 'micro'processor.
Technical matters aside, I guess that'd be like releasing your sophmore album when everyone is still grooving to the first one. I think people have a limit - a measurable one - to how quickly they'll bounce to the Next Best Thing.
So what I'm saying is there might very well be market disincentives for doing such things. IANAE (I Am Not An Economist) and I can't prove it, just taking a wild swing.
.02
My
Quux26
My
Quux26
www.crashspace.net
I certainly did not make those numbers up -- they were exactly what was taught in high-school science classes -- particularly, chemistry, I think. (It was written in the textbook, too.) It was a long time ago, so I obviously can't give a complete bibliographic refference (though it shared the title of "Chemistry" with virtually every other such textbook). I am certain I remeber the numbers correctly, despite the time elapsed, though it is possible that author refferenced them to cm rather tham m, for some cheesy reason (like assuming this would seem small to "kids"); E -6 m = E -4 cm, E -10 m = E -8 cm, so that would work.
"My" notation, is the "engineering notation" found on most calculators, BTW -- looking at the bottom of the screen reveals that <SUP> is not listed as available HTML codes, so I used this notation rather than exponents.
Note that I wouldn't have given them if I didn't have reason to think those were accurate. It is interesting, that some other replies managed to be informative, rather than just accusational and insulting.
You're right, its the transistors that (should) double not clock speed.
Thanks for that informative post!
The only reason all cover-ups appear to fail is that you never hear about the ones that succeed.
I work for a silicon wafer manufacturer, we supply Intel, AMD, etc with the wafers that they put the chips onto.
.13 micron. The human eye can only see down to .30 microns or so. (That's under ideal conditions, dark room, 200 watt halogen light focused on the wafer.) Needless to say there has been a transition to machine only inspection as the chip lines widths have gotten smaller.
.13 micron line widths makes me wonder what the yields will be like. The machines that can see down to .13 micron run about 1/2 million or so. I haven't seen what the next generation of them will do, but I know that we don't have them installed in our plant yet. That is definately one reason why it takes so long for this stuff to reach consumers. The clean rooms that the wafers are cleaned and inspected in are filtered down to class 10, which translates to less than 10 particles in the air at .3 microns. We usually have 0-1 at .1 microns. Makes an operating room look positively filthy.
The wafers we supply have and 'Epi' layer on them, which is short for epitaxial. The layer is silicon that is grown on the wafer at a high tempurature (I think 950-1000 degrees C). This makes the wafers less rough, thus smaller lines widths. The wafers are inspected for defects, and the machines that inspect them can only see particles, pits, etc down to
Intel already annoucing the
The wafer manufacturers are mostly breaking even at this point. Intel is making fat cash, but they are getting it from squeezing all the wafer suppliers. Some have dropped out of the business do to the lean conditions. Nobody really has enough money to buy equipment to make these wafers on a large scale, let only finding vendors that have equipment that meets those specs.
.01 micron, holy crap!
Think you've been reading too much sci fi...
~ppppppppö
sorry for being rude. I thought you were trolling.
I have noticed this to. Object Pascal (I used the Borland Delphi 5 compiler) is blazingly fast, much much faster than MS VC++, Borland C++Builder or gcc (though Borland C++Builder seems to be somewhat faster than MS VC++). Even plain C compilation with those compilers is way slower than Object Pascal compilation.
Someone can give an explanation?
This sig under construction. Please check back later.
A kettle with 10^8 Kelvins should work :).
(That's what princeton's Plasma Physics Lab can do ).
Mode (3) smart-aleck mode. Press * to return to main menu.
If this is to be a point on the Moore's law curve it will have to be in production in just over 6 years.
Can't have it both ways. Either it (or something of equivalent density) is out then or Moore's law finally breaks down.
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
I don't think so. Extrapolating Moore's law backward to 1930's, even with only a factor of 2 per 18 months, gives 1 operation per day. Somehow I think computations went a little faster than that, and certainly faster than 1 operation per 3 millenia in 1900.
If Alan Turing did indeed propose such a doubling every 12 to 18 months (and nothing in my reading suggests it), then such progress drastically slowed some time in the last 70 years.
-- Tim Little
I'm assuming that IBM would be tailoring this process towards MOS technology. Sure, they can make the channel length in a MOSFET really small, but wouldn't this introduce a lot of problems with the existing MOS model? One of the biggest problems with MOS technology is scaling dimensions down causes rampant power loss (due to leakage currents) and dominating electric field effects that can totally destroy transistor operation. You can't just go from 0.13 micron to 0.01 micron with the same basic structure and expect the same type of operation. Is there a new kind of transistor technology being proposed?
The problem with this and all new tech is cost. Right now, AFAIK the real research is going into .07 micron tech (you say .18, but actually .13 is "cutting edge"). Currently .07 is done using a laser etch, as opposed to lithography. For any new tech to become available you have to be able to mass produce it, and do so cheaply. .07 isn't cost effective yet (because of the laser it takes a long time to produce one wafer), and I would guess that .01 is not cost effective either. the next 10-15 years will likely see IBM first perfecting the process, and then scaling it to large scale/mass production. After that you may see things (5-10 years) being created using this process, but I wouldn't expect anything, even larger feature sizes, to be seen before then.
From first proof of concept to a commercialy viable product is often very very long. Such as:
Liquid fuel rockets - 1920's
Turbojet - mid 1930's
TV - 1920-something
High temp superconductor 1992?
Digital electronic computer - 1945
They haven't even made a chip yet! They have just made a transistor or two.
It is not even clear from the article how small they have gotten the channels: it only says that the technique "scales to" 10nm.
There is a long way from showing that a given technology will sort-of-work in a lab to mass-production. That's one of the reasons for the delay.
Another is that there might not be a market. It is currently quite feasible to get a couple of ~1GHz processors with a few gigs of RAM in a machin we can almost afford. Let's face it: few of us sitting here reading Slashdot are using our quad-Xeon workstations to their fullest. Who would really buy it at, say, 100 times the current price? I've just ordered my dual-PIII and I doubt I could easily use more processor speed. (Memory, maybe.) And I'm sure I wouldn't pay half a million bucks for it!
Hi!
Another problem would be IBM would have a proprietary production design and "closed-source" is evil (around here anyway)
It sucks, but that's capitalism.
Stop running the SETI@home! It's NSA's Echelon client and by running it, you're helping the Big Brother!
Just using the word is racist, and it's worse if you are indeed African-American. Everyday people like me are fighting to achieve equal standing with all other races, and here you are making yourself look like a fool. Way to represent bro.
I have one question here: will software really need more and more CPU performance as time goes by?
I mean, as the article says, sure, servers and stuff will definitely put good use to the increase in performance, but what about good ol' Joe Sixpack using Excel at his office? I mean, besides from cranking SETI@home units faster, is there really such a need for faster processors at home / office?
In all honestly, I stopped noticing any speed differences around 200MHz or so. I used a 200MHz Pentium running Win NT for a while at work, then I went to a 400MHz Pentium II. Couldn't tell the difference at all.
It is getting to where rewriting software and/or changing your approach are much more valuable than processor pissing contests. When I compile code with Visual C++ it seems to take forever, given a large project. If I use Object Pascal instead, the compilation time drops by 2-3 orders of magnitude. That's a much bigger win than increasing my machine to a 2GHz processor.
As circuit density increases, stray radiation from everyday household sources may become a serious source of errors in computers. Radioactive thorium can be found in gas latern mantles and in high quality camera optic lenses. Amercium can be found in smoke detectors. Even all potassium is mildly radioactive (decays into argon, hence the potassium-argon dating method). Everyday carbon-14. We may need to start having lead shielded PC cases. Of course the future Imacs will use lead crystal with various impurities to make translucient colors.
my comment here is a bit late, but yes, it was sarcasm on the IBM doesn't want to make intel mad and notice i said "mass market"- theres a market, but the average computer user won't be needing this for a while.
And 640K ought to be enough for anyone
This breaktrough is just one of the many, many hurdles that have to be overcome, in order to get a product that can actually be manufatured.
from http://linuxtoday.com/stories/296.html:
"On somewhat of a tangent, there is continuing work to support a subset of the Linux kernel on 8086, 8088, 80186, and 80286 machines. This project will never integrate itself with Linux-proper but will provide an alternative Linux-subset operating system for these machines. "
I think that aswell as being a 16 bit chip another problem to porting linux to the 8088 was the memory. Was it that the 8088 didn't support protected memory? I forget.
You are right that this guy almost certainly has never never tried Linux on an 8088 but it's not imposible.
I would worry the salmon chip trying to swim upstream to spawn. The worst we have to feer from the V-Groove is some funkadelic dancing.
---
/bin/fortune | slashdotsig.sh
Well we arn't in the same domain. .. Ohhh nevermind..
As Newtons laws are accurate, but only the the domain of speeds not approaching c.
As Moore's laws are acurate, but as thing have definatly changed since the creation of Moore's law it is almost accurate to say we exist in a different domain.
Maybe with this tech 3dfx can make a voodoo5 small enough to fit inside my computer case.. :)
Ahem, a law can not be both accurate and in need of revision within its (original) domain.
Hi!
This is yet another "I've always wondered...." question I have for all you Slashdot readers.
When we see stories about quantum leaps in computer technology, why are companies so slow to actually produce, implement, and sell it?
I feel releasing this technology now would not only benefit consumers, but help to drive down prices of other technologies. For example, if IBM released a processor built using this process today, I'm confident Intel's CPU price would drop.
So, what's keeping IBM from releasing hardware based on this technology in 1 to 2 years instead of 10 to 15? Ideas?
.18 to .01 is quite a leap... but could this process be used today, though on a smaller scale? If it's going to take 10 to 15 years for production at .01, how about .10 in a year? Or would it be impractical or inefficient to do that?
It's interesting to think that if people weren't always thinking that Moore's law can't stand up to more than another 10 years, we'd need a new law.
What I mean is that since it takes about 10 years for an emerging technology to go from theory to mass implementation, if there were theories that showed the promise of Moore's Law living on for more than ten years into the future, products based on those theories would emerge faster than Moore's Law predicts.
Fox's Law: The estimated time that Moore's Law will hold true will always be close to the time it takes to turn the latest theory into a commercial product.
Kevin Fox
Kevin Fox
Quake Arena!
:)
In addition, IBM and Intel agree that, especially with faster Internet connections, software will catch up to and exceed the capabilities of today's desktop processors, requiring more performance there as well.
I have one question here: will software really need more and more CPU performance as time goes by? (Code it again, Sam!)
I mean, as the article says, sure, servers and stuff will definitely put good use to the increase in performance, but what about good ol' Joe Sixpack using Excel at his office? I mean, besides from cranking SETI@home units faster, is there really such a need for faster processors at home / office?
Shouldn't other areas of computer science be explored as well? Im sure there's lots of research going on all the time, but if someone were to discover a faster search / compression / whatever algorithm that would make up for a slower processor, wouldnt it?
As usual, that's my opinion... and as I said, the truth is I'll probably use it to play better, faster and bloodier games on my PC
Tongue-tied and twisted, just an earth-bound misfit, I
Learning to fly, Pink Floyd.
Ah yes - this is indeed true. However, let me clarify (I should have been more specific earlier).
What has been doubling over the last two hundred years isn't exactly computation in terms of operations/sec. It is actually computation in terms of operations/sec/price. Both Kurzweil and Minsky (and Turing) have written on this. If you trace the amount of computing power that $1,000 buys over the last hundred years or so - you can see that the amount of computation bought per $1k has been doubling every 12-18 months.
-=|t
Duh.
The moderation is really crazy nowadays. Bremsstrahlung is a physical process, which with high energy particles like X-rays can scatter the lattice of silicon, resulting in spurious irradiation and damage to components.
Mode (3) smart-aleck mode. Press * to return to main menu.
Last time I checked, a Micron was different from a micrometer. Specifically, a micron was E -4 (0.00001) Meters, 100 micrometers, or 0.1 milimeters. A micrometer was E -6 m, An angstrom was E -8 m, and a nanometer was E -9 m. Thus, 10 nanometers would be precisely one angstrom, but actually, 0.0001 microns.
Whose in error? Have I just been uninformed all these years, or is this confusion on the poster part, and a mistake in the original report / news release?
If you go any smaller, your waves become X-rays and that significantly complicates matters.
--
Groove-V. It has a certain ring to it.
Donate background CPU time to fight cancer.
Where did that number comes from?
The REAL fundamental is Mr Heisenberg's Uncertainty Principle pxh!
Mode (3) smart-aleck mode. Press * to return to main menu.
Assuming a doubling in clock speed every 2 years, in 20 years clock speed will increase by 2^10 ~ 1000. So PERFORMANCE should increase by a factor of 1000.
Obviously this is going to happen somehow but its nice to see how it might be done.
"In our days we only had 50GB hard drives and 2GB of RAM..... And we liked it!"
The only reason all cover-ups appear to fail is that you never hear about the ones that succeed.
With chips like this, we can fear only one thing: the marketing campaign. Intel's was bad enough, even without a name like V-Groove.
--Jeff
new option should be:- -1 Inciteful.
With a .01 micron pathway, strange quantum behavior would start taking over. IBM researchers are going to have to find a way arround the quantum effects that would render such a chip useless.
--
--
tobam'i: foo for the masses.
The article does not state they will be using this in 10-15 years in CPUs, but that they will be doing it as soon as the engineers figure out how t oapply the technology. Someone mis-read the text.
The article claims this technology will put IBM 10-15 years ahead of the Moore's Law curve.
To quote the article....
"...will allow the company to stay ahead of the curve of Moore's Law 15 to 20 years in the future..."
Yes it isn't the best way to express the idea but it's not that hard to understand.
This is why I don't read SlashDot very often. this is about the 10th time I've seen you mis-report an artice. I'm amazed at how many people respond to them without actually reading the article themselves.
Article X: The powers not delegated... by the Constitution...are reserved...to the people
yes, anisotropic etches are possible. They happen all the time...crystal fault delineation, dopant preferential etching.
I haven't read the article though, so maybe in their context it is wrong...
cheese
We've all heard a lot of talk about the "fundamental boundaries of silicon" and how "Moore's law will be destroyed in 15-20 years". While there are definitive boundaries of the material, the idea that computational speeds will taper off in 15-20 years is complete rubbish.
Most people know Gordon Moore's great "law". What most people don't know is that his profound statement of computation, a) isn't that profound b) isn't originally his idea.
Anyone that has read the works of our favorite British geek, big Alan Turing, knows that he stated in the 30s that computational speeds double every 12-18 months. Turing took a look at the "computers" dating back into the 1800s. From purely human/mechanical, to entirely mechanical, to electro mechanical, to electrical (Vacuum), to transister, to IC, computational speeds have been doubling since Babbage's girlfriend was writing theoretical software! (ok maybe a little later than that..)
The point is that at each stage in computing history, when one medium reached its limit, another picked up and continued seemlessly along. So the broader Moore's law (the smaller scale and actual Moore statement was only regarding ICs), will continue, silicon or no silicon. So then the interesting question is what will pick up when silicon dies. Molecular/Nano would probably be most peoples guess right now.
I'd expect the only failure of "Moore's law" to be that it underestimates the speed in which computing technology will double - my guess is that in 15-20 years it will go even faster than 12-18 months...
-=|t
Well, as far as I have read in most computer magazines, the Mhz is higher in the AMD and Inel chips, the G4 optimizes on it more efficiently.
/. in fact.
I'm not sure how accurate that data is, but I am sure I have read it multiple times, probably on
By the time Intel starts to manufacture these, I'm sure all of the othe rprocesor companies will be too. The Macs will inevitably benefit form this technology also, maybe you'll just have to wait a quarter though.
Twitter.com/TrentonHyatt
No, the 80286 supported protected mode, but the 80386 is so much of an improvement that nobody codes for the 80286. The 80286 had a memory limit of 16 Meg. Transition from protected to real mode via an aborted reset. The 386 also introduced the virtual 86 mode.