NYT: Computer Chips Seem Poised to Shrink Again
Premise Checker
http://www.nytimes.com/2010/08/31/science/31compute.html
[Reader comments added.]
Advances Offer Path to Further Shrink Computer Chips
By JOHN MARKOFF
Scientists at Rice University and Hewlett-Packard are reporting this
week that they can overcome a fundamental barrier to the continued
rapid miniaturization of computer memory that has been the basis for
the consumer electronics revolution.
In recent years the limits of physics and finance faced by chip makers
had loomed so large that experts feared a slowdown in the pace of
miniaturization that would act like a brake on the ability to pack ever
more power into ever smaller devices like laptops, smartphones and
digital cameras.
But the new announcements, along with competing technologies being
pursued by companies like IBM and Intel, offer hope that the brake will
not be applied any time soon.
In one of the two new developments, Rice researchers are reporting in
Nano Letters, a journal of the American Chemical Society, that they
have succeeded in building reliable small digital switches--an
essential part of computer memory--that could shrink to a
significantly smaller scale than is possible using conventional
methods.
More important, the advance is based on silicon oxide, one of the basic
building blocks of today's chip industry, thus easing a move toward
commercialization. The scientists said that PrivaTran, a Texas startup
company, has made experimental chips using the technique that can store
and retrieve information.
These chips store only 1,000 bits of data, but if the new technology
fulfills the promise its inventors see, single chips that store as much
as today's highest capacity disk drives could be possible in five
years. The new method involves filaments as thin as five nanometers in
width--thinner than what the industry hopes to achieve by the end of
the decade using standard techniques. The initial discovery was made by
Jun Yao, a graduate researcher at Rice. Mr. Yao said he stumbled on the
switch by accident.
Separately, H.P. is to announce on Tuesday that it will enter into a
commercial partnership with a major semiconductor company to produce a
related technology that also has the potential of pushing computer data
storage to astronomical densities in the next decade. H.P. and the Rice
scientists are making what are called memristors, or memory resistors,
switches that retain information without a source of power.
"There are a lot of new technologies pawing for attention," said
Richard Doherty, president of the Envisioneering Group, a consumer
electronics market research company in Seaford, N.Y. "When you get down
to these scales, you're talking about the ability to store hundreds of
movies on a single chip."
The announcements are significant in part because they indicate that
the chip industry may find a way to preserve the validity of Moore's
Law. Formulated in 1965 by Gordon Moore, a co-founder of Intel, the law
is an observation that the industry has the ability to roughly double
the number of transistors that can be printed on a wafer of silicon
every 18 months.
That has been the basis for vast improvements in technological and
economic capacities in the past four and a half decades. But industry
consensus had shifted in recent years to a widespread belief that the
end of physical progress in shrinking the size modern semiconductors
was imminent. Chip makers are now confronted by such severe physical
and financial challenges that they are spending $4 billion or more for
each new advanced chip-making factory.
I.B.M., Intel and other companies are already pursuing a competing
technology called phase-change memory, which uses heat to transform a
glassy material from an amorphous state to a crystalline one and back.
Phase-change memory has been the most promising technology for
so-called flash chips, which retain information after power is switched
off.
The flash memory industry has used a number of approaches to keep up
with Moore's law without having a new technology. But it is as if the
industry has been speeding toward a wall, without a way to get over it.
To keep up speed on the way to the wall, the industry has begun
building three-dimensional chips by stacking circuits on top of one
another to increase densities. It has also found ways to get single
transistors to store more information. But these methods would not be
enough in the long run.
The new technology being pursued by H.P. and Rice is thought to be a
dark horse by industry powerhouses like Intel, I.B.M., Numonyx and
Samsung. Researchers at those competing companies said that the
phenomenon exploited by the Rice scientists had been seen in the
literature as early as the 1960s.
"This is something that I.B.M. studied before and which is still in the
research stage," said Charles Lam, an I.B.M. specialist in
semiconductor memories.
H.P. has for several years been making claims that its memristor
technology can compete with traditional transistors, but the company
will report this week that it is now more confident that its technology
can compete commercially in the future.
In contrast, the Rice advance must still be proved. Acknowledging that
researchers must overcome skepticism because silicon oxide has been
known as an insulator by the industry until now, Jim Tour, a
nanomaterials specialist at Rice said he believed the industry would
have to look seriously at the research team's new approach.
"It's a hard sell, because at first it's obvious it won't work," he
said. "But my hope is that this is so simple they will have to put it
in their portfolio to explore."
---
Advances Offer Path to Further Shrink Computer Chips--Readers' Comments
http://community.nytimes.com/comments/www.nytimes.com/2010/08/31/science/31compute.html
1.
Larry Fine
NYC
August 31st, 2010
8:15 amThere should be more government investment in this research, because in
the future they can put these small devices in the general populations
brains. This way, the people can be programmed to think the right way.
Recommended by 44 Readers
2.
MS
New Jersey
August 31st, 2010
8:31 amI remember when there was talk of the possibility of there someday
being a one megabyte storage disk.
Recommended by 14 Readers
3.
RKK
NY
August 31st, 2010
8:31 amSounds interesting... would like to read more on the technological
viability of the memoristor project. Does heat travel faster than
electrons in phase-change memory ?
Recommended by 3 Readers
4.
GSH
RI
August 31st, 2010
8:32 amClassical hype. Nothing concrete behind it. Also, the reporter does not
seem to understand the difference between memory and logic. What is not
keeping up with Moore's law is logic, namely ever faster switches. You
can shrink memory all you want, if logic speed fails to improve,
computer will not become faster.
Recommended by 42 Readers
5.
chris
new york
August 31st, 2010
8:40 amI bet China has millions and millions of smart people like Jun Yao.
Give them the right resources and they will unleash the Next Big
Things. Wait in the next decade or two. China could be the next
powerhouse for scientific discoveries.
Meanwhile, back in America, our students still think it's cooler to be
a dumb athlete than a science wiz.
Recommended by 59 Readers
6.
Zoe
NJ
August 31st, 2010
8:48 amYeah this is great news, barring any misuse as wisely mentioned in
Larry Fine's astute comment (#1). We may have very small light personal
computer/communication devices. Sadly, the manufacturing and related
jobs will not occur in the United States.
Recommended by 31 Readers
7.
EHS
Brooklyn
August 31st, 2010
8:48 amWonderful. Now let's shrink the number of Americans who die of cancer
every year.
Recommended by 16 Readers
8.
Eli
Boston, MA
August 31st, 2010
8:55 amIf I was a dirty energy business I would be trebling and moving out of
coal and oil as fast as I could.
These new advances have huge implications for the generation of clean
and healthy energy from sources that are forever such as sun and wind.
Recommended by 14 Readers
9.
Barnegat Leight
Beach Haven, NJ
August 31st, 2010
9:00 amThis technology has been around awhile.
HP 'announces' it out every now & then.
More power to them.
http://www.nature.com...
Found: the missing circuit element
Described in 1971, made in 2008:
'memristors' promise a computer revolution.
High-school physics students grappling with the delights of capacitors,
inductors and resistors will be groaning into their exercise books.
Electronics experts in California have finally succeeded in proving the
existence of a fourth fundamental unit of electronic circuits: the
'memristor'.
The existence of the memristor, short for 'memory resistor', was first
suggested in 1971, but only now have researchers succeeded in creating
a real, working example. They hope that the new components could
revolutionize computing, promising an end to frustrating waits for your
computer to boot up.... (April 30, 2008)
Recommended by 12 Readers
10.
Ken C. A.
Los Angeles
August 31st, 2010
9:05 amThese new technologies just take us one step closer to day when AI
takes command. The next step in the evolution in the area of the
universal niche of intelligence is not biological but it is machine.
Will these ultra intelligence self motivated machines find it necessary
and logical to keep us around, probably not. Astronomical increases in
AI is a forgone conclusion. Yet, with luck they may find it useful to
move on to other regions of the universe as they seek higher forms of
knowledge and maybe even one day they will make their own universe that
would better suit them. In the begininng they will be used as simple
pets and toys and next they will be used as servants and employees,
soldiers and that sort of thing though finally they will govern all
life on earth. Ken C. Arnold Santa Monica, Ca
Recommended by 7 Readers
11.
Llyn
Wisconsin
August 31st, 2010
9:15 amI'm sure that smaller chips with more memory will be useful in many
devices, but the electronics industry needs to realize that the drive
to make devices used daily by people smaller and smaller has more than
reached its limit. While young people with 20/20 vision may be able to
read tiny text and tiny labels on tiny buttons on tiny devices, older
people (and the vision impaired) are not nearly as able to read and use
these devices. Finding them a such a struggle, we are no longer part of
the market for them. Get real.
Recommended by 12 Readers
12.
J
Northeast
August 31st, 2010
9:40 amSo I see the implications for storage and for us eggheads, that is
interesting. But, the push for faster and faster chips for consumer
PC's has always baffled me, given that most people use their computers
for Internet and email. Could the time/money be better utilized
elsewhere?
Recommended by 11 Readers
13.
Steve
Hull MA
August 31st, 2010
9:40 amIn the late 1970s I purchased my first personal computer, before the
IBM PC was invented. There were no applications. If you wanted it to do
anything you had to program it. If the memory in my present computer
cost what the memory in that one did, the present computer would have
cost several hundred million dollars.
So what? What fundamental cultural advances do we have to show for it?
Recommended by 15 Readers
14.
Ben
Austin
August 31st, 2010
9:45 amNow when will we find technology to have similar effect on the federal
budget deficits?
Recommended by 5 Readers
15.
Judd
New York
August 31st, 2010
9:55 amThis advance, like most others in computing signals the increased
ability of computing to be involved in more aspects of our daily lives.
For instance, I am surprised that no one is recognizing how much we use
computing in our cell phones and mobile devices. We also use computing
increasingly in reading devices, in household devices of all sorts,
extensively in our cars, workplaces and public places.
The question in my mind is will we implement advancing computing with
enough foresight to avoid and reverse extensive environmental
degradation and social alienation? Based on our recent track record,
the answer is no.
Recommended by 4 Readers
16.
YL
New York, NY
August 31st, 2010
10:08 amI do not see the article in Nano Letters fast track publication (the
current, regular issue is from August 11th):
http://pubs.acs.org/toc/nalefd/0/0
It is usually a bad style to making press releases before the
publication of the actual paper. As already commented by other people
here, journalists are not researchers in the field and typically cannot
properly review the claim.
Recommended by 6 Readers
17.
Ben
New York, N.Y
August 31st, 2010
10:09 am@RKK You might be interested in this talk at HP Labs which discusses
the technical aspects of memristor technology:
http://www.youtube.com/watch?v=bKGhvKyjgLY
Also, in phase change memory, the scales are very small and the heat
changes are initiated by electrons, so the speed is compareable or
better then current NAND type flash memories.
@GSG I think the article might have glossed over some details of these
technologies to make them understandable to the public, but as a
computer engineer myself, I assure you that a claim like 'nothing
concrete behind it' demonstrates a great ignorance about the subject.
Firstly, Moores Law says nothing about logic, just transistor density.
Secondly, there are many architectural approaches that can be taken in
modern chip design to increase apparent speed to the user (hopefully
you would have heard of instruction pipelines and other basic
micro-architecture approaches?) Additionally, modern CPU's like Intel's
i3, i5 and i7 series are examples of this, as is GPGPU computing, in
which clock speeds have remained relatively steady (or even decreased!)
but effective computational power has increased exponentially.
Finally, research into raw transistor speed has demonstrated logic
operating at over 500 GHZ effectively (IBM). You need only google the
terms "transistor speed" to see many examples of this.
Recommended by 9 Readers
18.
YL
New York, NY
August 31st, 2010
10:09 am#5:
"Meanwhile, back in America, our students still think it's cooler to be
a dumb athlete than a science wiz."
This is only a natural consequence of NCAA and the availability of
athletic scholarships, which is unheard of in most countries including
China. We would not have this craze, if all schools are held to the
prohibition of athletic scholarships, like the Ivy League.
Recommended by 16 Readers
19.
boatbutter
St. Louis
August 31st, 2010
10:09 amTruly remarkable. Can this technology be applied to the egos of
politicians?
Recommended by 2 Readers
20.
Aaron H.
OH
August 31st, 2010
10:10 amAllow me to sum up the article for everyone in a lengthy headline:
"New Untested and Unverified Computer Advancement that I Cannot Even
Explain in Any Detail May Change the Face of Computers Forever; Moore's
Law Still Makes Sense"
NYTimes is reading like Slashdot this morning I guess. No offense
intended but I got really excited by this article, then quickly
realized that the reporter wasn't going to make any serious attempt to
explain what he's talking about in any sort of real detail. Add in the
obligatory Phase Change Memory paragraph and you've got a front page
article!
Recommended by 5 Readers
21.
Haz
Illinois
August 31st, 2010
10:10 amGreat... For China. Could it be possible that these chips will be
manufactured in the USA? Not a chance.
Recommended by 15 Readers
22.
markhas
Whiskysconsin
August 31st, 2010
10:10 amEli! the sun is not a permanent source, it is finite and will burn
itself out, just wait awhile you'll see.
Recommended by 3 Readers
23.
Arun Mehta
New York, NY
August 31st, 2010
10:13 amWaiting for the day that chip will shrink the human ego to it's real
worth.
Recommended by 4 Readers
24.
Larry Mark
New York
August 31st, 2010
10:13 amI believe in the reverse of Moore's law.... an observation that my
Weight doubles every 18 months.
Recommended by 5 Readers
25.
Jersey Wonderfool
Princeton Junction, NJ
August 31st, 2010
10:13 amHere we go again. The real pin head sized brain! Now only if we can
miniaturized politicians mouths and ......
Recommended by 4 Readers
26.
Phil
Brooklyn
August 31st, 2010
10:13 amSteve, as you were typing your comment so that all of the readers of
one of the world's largest newspapers could read your opinion, could
you really not think of any cultural advances the development of chip
technology has brought about?
Recommended by 7 Readers
27.
Jason Wolffe
Washington D.C.
August 31st, 2010
10:25 amThis is no surprise. They have been developing this new chip technology
for a few years now and have discovered a whole new level of "Nano"
circuitry. Bravo for the scientists!
I was seeking semi-skilled full time employment with a busy high tech
company in Bedford Massachusetts, only to find that all of the product
manufacturing is done in India, and Indonesia. Any advances discovered
here should be manufactured here.
What part about "Unemployment in America" do these people not
understand?
Recommended by 11 Readers
28.
Ed Burke
Long Island, NY
August 31st, 2010
10:30 amMaterials we are all familiar with like metals, cellulose, silicone,
and other natural materials behave very differently at the Nano level.
A Coating of nano fibers made with the pigment titanium is transparent
at the nano level, yet opaque as a pigment in paint. The idea of using
nano fibers to make complex electronic circuits has reached its point
of relevance. These new ultra small circuits hold the promise of less
energy to operate, less heat energy to dissipate, and as this article
says, vastly greater amounts of storage space.
God in his wisdom, is providing his creatures with more of his
knowledge, yet again. It will be mankind's obligation to use this
knowledge to serve God, and do good, as opposed to what mankind usually
does with God's gifts, use them to do evil.
Recommended by 2 Readers
29.
Elephant lover
New Mexico
August 31st, 2010
10:50 amI am delighted that the US still has the power to innovate. I agree
with Haz that it is unfortunate that making these chips will
undoubtedly be done overseas.
Recommended by 8 Readers
30.
Kodali
VA
August 31st, 2010
10:55 amThere are very good reasons why people are skeptic of the research
reports. The material behavior at nano scale is not completely
understood. It is important to focus on understanding the material
behvior at those scales before a reliable products for commercial use
can be developed. The industry is investing heavily and the federal
government should provide additional funds to keep America ahead of the
rest of the world. Our economy and our future depends on technological
advances and we need the research scientists who are struggling in the
labs for small amount of salary while Wall Street bonuse runs in
billions of dollars for the work that has no value. We have to thank
all those research scientists.
Recommended by 6 Readers
31.
Tom
Phoenix
August 31st, 2010
11:20 am#21 "Great... For China. Could it be possible that these chips will be
manufactured in the USA? Not a chance."
This is a bit off topic but Intel and others do have fabs in the US, so
there is a good chance that the latest generation chips would be
manufactured here (to be fair Intel does have fabs overseas too).
But your comment speaks to the larger issue of the loss of
manufacturing in the US. That outcome can be put squarely at the door
of US policy makers who permit the importation of goods into the US
without tariffs. Consider that Germany has a balance of trade with
Asian countries because they protect, through tariffs, their
industries. Simple as that.
Recommended by 5 Readers
32.
Peter N. Budzilovich
Nyack, NY
August 31st, 2010
11:40 amPerhaps it's time to learn data-packing techniques from the Creator--
He packs data in DNA cells at 10-to-the-21-power bits/cubic centimeter.
In modern microchips, data density is on the order of
10-to-the-7th-power bits/cubic centimeter. I.e., lower by 13 orders of
magnitude! To appreciate these numbers, let's recall that the
information in all worlds' libraries is estimated at
10-to-the-18th-power bits. If these data were recorded in DNA
molecules, they would fill a volume of about 1% of an ordinary pin
head. If microchips were used, it would take a stack of chips piled up
from the Earth surface to the Moon! How is this for the Intelligent
Design, regardless what Charles Darwin says?
Recommended by 2 Readers
33.
westwing71
Minneapolis
August 31st, 2010
11:40 amIn the IEEE journal article related to memristors--published a few
months ago--the statement was made that it was considered possible
that the nature of the technology might allow it to be used to develop
machine intelligence. Supposedly this related to characteristics that
mimicked neurons, however, my thinking on the issue is that individual
neurons have to combine in a special way to produce intelligence and
the memristor will not be able to duplicate that. On the other hand, it
could easily shrink the size of the memory element further, as well as
storing more data using one memristor... and that would provide a
massive increase in the amount of data which could be packed into a
small space.
Recommended by 1 Reader
34.
Domingo Tavella
San Francisco
August 31st, 2010
12:03 pm
These advances also reveal an unbelievable degree of cultural
schizophrenia. While scientific breakthroughs in the US are carried out
by atheists, Hindus and Jews, the thrust of the American culture
becomes ever more Islamic-like in its Christian dogmatism. Perhaps
scientists are too busy with their work to notice that their way of
thinking is increasingly less welcome in the land of superstition.
Recommended by 6 Readers
35.
Mauren
Holden, MA
August 31st, 2010
12:03 pm
For a headline story, this was awfully thin on facts and no actual
breakthroughs are reported. The headline led me to believe otherwise.
Recommended by 2 Readers
36.
Out with a whimper
New York City
August 31st, 2010
12:03 pm
Just what the world needs. Up their nose with a rubber hose.
Recommended by 0 Readers
37.
akka
Paris
August 31st, 2010
12:05 pm
I wonder if a scientist like Moore ever tried to figure out how much
memory it would take to store all one's memories at the end of his
life.
I'm sure we are not far from the point where we can carry small
electronic devices, plugged at an early age, that would record all our
conversations the things we hear and see during our lives.
What's the limit? 1 Terabyte? 1000 Tb? or maybe 1000 000?
Recommended by 0 Readers
38.
mandragola
In the OC, CA smog...
August 31st, 2010
12:15 pm
Have we invented a bubble and bonus chip yet so that the geniuses on
Wall St. will become to small to fail?
Recommended by 0 Readers
39.
SM272147
AZ
August 31st, 2010
12:15 pm
There are many other technologies that promise to keep Moore's law
alive into the future. But the problem is not shrinking the transistor,
but how to cool these chips, you don't want to carry a fan on your cell
phone, do you? This article sheds no light on that fundamental hurdle.
Recommended by 0 Readers
40.
Steven
Washington DC
August 31st, 2010
12:15 pm
@5 chris, yeah, but the thing is, their millions of smart kids never
reach their potential until they come to a free society like the good
'ol USA. I'm not worried about China ever passing us until they have
political reform along with economic reform.
Recommended by 0 Readers
41.
mandragola
In the OC, CA smog...
August 31st, 2010
12:15 pm
Have we invented a bubble-and-bonus chip yet so that the geniuses on
Wall St. will become too-small-to-fail?
Recommended by 0 Readers
42.
Scott
Middletown, Ohio
August 31st, 2010
12:15 pm
It is a good thing that carbon paper is hard to find, because I thought
this was a copy of several articles I have read in the last couple of
years.
I am waiting on Quantum memory look that up and give us an update, this
is old news.
Recommended by 0 Readers
43.
Adil
NY, NY
August 31st, 2010
12:15 pm
I find it rather wasteful that people are spending so much to speed
hardware when the problem lies in the software.
yes the speed of hardware is very very fast enough... thats why
hardwired programs can run anything very reliably--from complex
routers to rockets (rockets are easy to control I suppose).
but that doesn't mean that any IBM or MS software will run faster...
just take the browsers--I have noticed that it is becoming slower in
last few years... some softwares (java IDEs esp) have improved speed..
but when you look at the possibility of speed, software is a big big
bottleneck.
why there is no reseach about:
#1. measuring performance of softwares in numbers (just like we measure
very complex car with simple numbers which even layman can understand)
#2. why we never think about certification of softwares so as to weed
out bad apples (belive me there are tonnes of them) from software
industries??? in fact, bad programmars are actually rewarded with more
work in support.!
#3. why we don't research about how software ages? like even a layman
can tell age of car with simple number on speedo.. why it is so hard to
tell age of softwares?
All I can say is that all these might be gr8 only for storing devices
which are becoming redundant anyway.. like I used to buy DVDs in
2004s.. but now I just use streaming. who needs bigger capacity chips?
Recommended by 0 Readers
44.
Whatever
nh
August 31st, 2010
12:15 pm
Rather than constantly whine, people should perhaps think a little
harder about--and address head-on--**why** manufacturing has left
the US in droves. If you think it's just about direct labor costs,
think again.
Look all around you, at the quality of nuts-and-bolts labor.... whether
working in factories, building tunnels and bridges, serving as a
bureaucrat, running a telecom or cable business, serving behind a
counter at McDonalds/Walmart, or fixing your plumbing, and ask, might
we have a quality problem in this country?
Hint: Where do you think the parts that got Toyota into trouble over
stuck accelerator pedals got made?
We're (thank God) still great at innovating, but not necessarily
implementing.
Recommended by 1 Reader
45.
A Smith
DC
August 31st, 2010
12:15 pm
So when are we gonna get an eyePhone? (See: Futurama)
Recommended by 0 Readers
46.
GSH
RI
August 31st, 2010
12:15 pm
Ben, I happened to lead a team in the late '80s, when we first broke
100GHz with silicon devices. I happen to know more about computer
architectures than you can possibly imagine. Pipelining reached it
limits, massively parallel computing show no sign of breaktrough. (Non
even remotely) Be assured, it is nothing but hype. And, BTW, there is
no natural law out there that progress in science, engineering, etc.
has to continue. Get used to it, we are stuck.
Recommended by 0 Readers
47.
mike
calabash , n.c.
August 31st, 2010
12:15 pm
What a bunch of malarky! Just show me how i can turn the channels on my
tv without haveing to get up out of my chair so often. Colored TV? What
is that going to prove. Arnt they pushing it a bit? Is it small enough
so as you can see it? I dont get it!
Recommended by 1 Reader
48.
Rlanni
Princeton, NJ
August 31st, 2010
12:20 pm
Just more computer power and memory for ever more bloated operating
systems and applications. If only the software end had a Moore's law,
not to mention, batteries.
Recommended by 0 Readers
49.
Einstein
America
August 31st, 2010
1:32 pm
The maxim still remains: Garbage in, Garbage out.
More important than speed or capacity is who will be programming the
apps? For what purpose--for the good of humanity or merely to control
humanity?
Recommended by 0 Readers
50.
larry
Amissville, VA
August 31st, 2010
1:32 pm
Great. Now there will be more resources for Microsoft et al to consume
without adding any real benefit to the user.
Recommended by 0 Readers