Tech Talk iconTech Talk

Exascale supercomputers: Can't get there from here?

Today Darpa released a report I've been hearing about for months concerning whether and how we could make the next big leap in supercomputing: exascale computing, a 1000x increase over today's machines. Darpa was particularly interested in whether it could be done by 2015.

With regard to whether it could be done by 2015, the answer, according to my read of the executive summary, is a qualified no.

In it's own words, here's what the study was after:

The objectives given the study were to understand the course of mainstream computing technology, and determine whether or not it would allow a 1,000X increase in the computational capabilities of computing systems by the 2015 time frame. If current technology trends were deemed as not capable of permitting such increases, then the study was also charged with identifying where were the major challenges, and in what areas may additional targeted research lay the groundwork for overcoming them.

The study was led by Peter Kogge, an IEEE Fellow and professor of computer science and engineering at Notre Dame University. (We'll be talking to him next week about the study for further coverage in IEEE Spectrum) And it had contributions from some of the profession's leading lights including Stanford's William Dally, HP's Stanley Williams, Micron's Dean Klein, Stanford's Kunle Olukotun, Georgia Tech's Rao Tumala, Intel's Jim Held and Katherine Yeolick (who I include in this list not because I know who she is, but because she lectured about the "Berkeley Dwarfs").

Darpa's helpers seem to have come to the decision that current technology trends will not allow for exascale computing. That's summed up pretty neatly in this graph, which clearly shows that the trend line in computer performance undershoots exascale in 2015 by an appreciable amount:

exascaleGflops.gif

The group found four areas where "current technology trends are simply insufficient" to get to exascale. The first and what they deemed the most pervasive was energy and power. The Darpa group was unable to come up with any combination of mature technologies that could deliver exascale performance at a reasonable power level:

exascaleGflopswatt.gif

The key, they found, is the power needed not to compute but to move data. Data needs to move on interconnects and they found that even using some really cool emerging technology it still cost 1-3 picojoules for a bit to go through just one interconnect level (like from chip to board or board to rack). Scale that up and you're talking 10-30 MW (167 000 - 500 000 60 watt light bulbs) per level. Eeesh.

The other 3 problems are memory storage (how to handle 1 billion 1GB DRAM chips), concurrency and locality (how to write a program that can handle a billion threads at once), and resiliency (how to prevent and recover from crashes).

These are equally interesting, but the power problem is, I think, what much of today's computing work is really boiling down to. Solve that, and things will look a lot sunnier for everything from high performance computing to embedded sensors.

The full (297 page) Darpa Exascale Computing report is here.

(In the November issue of IEEE Spectrum, watch for a cool simulation that Sandia computer architects did to show another bump in the road to future supercomputers. Their simulations show that as the multicore phenomenon advances in the processor industry, some very important applications will start performing worse.)

Nuclear waste imports can wait

Last July, our Sally Adee, brought you a story on the controversy over a Utah company's plan to import 18 000 metric tons of Italian nuclear waste into the United States and (after some difficult to understand process) dump some of it in Utah.

The Wall Street Journal reports that the Nuclear Regulatory Commission has decided to delay its decision on whether or not the importation can proceed. The NRC is going to sit on its hands until a federal court hears a related caseâ''some time next year.

The delay, says the Journal, gives a boost to a bill that would ban nuclear waste imports (unless they were defense-related). The legislation is currently stuck in committee.

Physics Nobel for why the Big Bang wasn't a big bust

From our intrepid intern, Monica Heger:

The Nobel Prize in physics was awarded today for discoveries in subatomic physics. Yoichiro Nambu, from the Enrico Fermi Institute at the University of Chicago won half the award for his discovery of the mechanism of spontaneous broken symmetry in subatomic physics. Two Japanese physicists, Makoto Kobayashi from the High Energy Accelerator Research Organization and Toshihide Maskawa from the Yukawa Institute for Theoretical Physics at Kyoto University, split the other half of the award for their discovery of the origin of broken symmetry, which predicts the existence of at least three families of quarks, a fundamental particle.

Broken symmetry lies behind the very nature of our existence. At the time of the Big Bang, if equal amounts of matter and antimatter were created, they theoretically would have destroyed each other. Instead, that symmetry was broken, allowing for the existence of our universe. Scientists still do not know how that symmetry was broken.

The three Nobel winners all explained broken symmetry within the framework of the existing laws of physics. Kobayashi and Maskawa were only able to do this by expanding broken symmetry to include three new families of quarks. The quarks they described in 1972 have only recently been observed in laboratories by particle accelerators.

Marching to the Beat of a Different Drummer in Nanotech

Andrew Maynard in his latest blog site presents one of the stronger metaphors I have seen to date to describe the state of dialogue (or lack thereof) on the future and direction of nanotech.

Maynard likens the current discourse to the latest social phenomenon the â''silent raveâ'' in which everyone shows up at the same place but listen to their own iPod.

These nanotechnology meetings to which Maynard draws his comparison consist of scientists, policy makers, industry leaders and NGOs just to name the main groups and they are all marching to the beat of different drummers.

What Maynard seems loathe to point out is that there may actually be qualitative difference between the drummers, or, to follow his metaphor, songs. Maybe Ringo Starr was a better drummer than Pete Best.

After reading TNTLogâ''s recent experience at another stakeholder consultation group intended to be â''Fruitful Dialogueâ'', one wonders how fruitful these dialogues can be when one or more of the groups clearly have absolutely no idea of what they are talking about.

Is it possible to step in and pull the plug on those iPods of the clearly misinformed? Probably not. The thought that some ideas and opinions are just bogus has come to be so anathema to â''reasonableâ'' people that we have to endure nonsense, or noise, and hope that the more pleasant notes come to the fore.

Unfortunately, hoping for something to happen doesnâ''t mean that it will.

Flash of Genius: See the Movie, then Read the Article

I watched Flash of Genius in a sort of slack-jawed amazement. The movie, which opened over the weekend, stars Greg Kinnear as Bob Kearns, inventor of the intermittent windshield wiper who brought car companies to their knees in the early 90s by winning a patent infringement case with GM.

The climactic scene is a patent trial! Sure, plenty of movies have trial scenes--but patent law is notoriously opaque. Throw in the complexities of engineering and a mentally disturbed engineer representing himself, and you've got the makings of cinematic Ambien. Mercifully, the trial moves at a brisk clip, with plenty of drama, and the most cogent explanation of the legal standard of non-obviousness that I've ever heard.

And of course, there was the flash of genius, Kearns' "eureka moment," when looking in the mirror and watching his eye blink, he realized he could make a windshield wiper work the same way. I was so excited when I came out of the movie that I started madly Twittering my review.

As I tweeted, I got to thinking about the nature of eureka moments. Kearns' eureka moment was actually several moments spread over a decade. Kearns blinded himself in one eye when popping a champagne cork on his wedding night. He and his wife recount this incident throughout the movie as the moment--but the wiper wasn't even a twinkle in his black eye at that time. Then there was the time he was driving his family in a downpour and he was frustrated by a lack of wiper-speed variability. Then there was the moment when he looks at his eye blinking in the mirror, and the first two moments came together--Eureka! Kearns' flash of genius.

Hmm. Flashes of genius, maybe. Or flashes that result in a moment of insight.

But that's really kind of nitpicky. Less so is the scene where Kearns' wiper works--the first time he turns it on. No way, I thought. And when I read John Seabrook's 1993 New Yorker article "The Flash of Genius" on which the movie was loosely (as it turns out) based, I learned that Kearns actually spent months perfecting his invention.

In fact, comparing the article to the movie is an object lesson in how Hollywood distorts complicated issues and complex people into a digestible package of entertainment. There's a laundry list of differences between legal fact and movie fiction. For instance, in the movie Kearns is a professor at Wayne State University teaching applied electrical engineering. In reality, at the time he invented the wiper, Kearns, who had a masters in mechanical engineering and would eventually teach at Wayne State, was commuting to Case Western Reserve in Cleveland, trying to earn his PhD. He came home on weekends to be with his wife and family. The real hero of the story should be his wife Phyllis, who had to take care of six kids by herself, while holding down a substitute teaching job (wait, was that true or made up?).

Finally, there's the explanation of the technology behind the intermittent wiper, or rather lack thereof. In the movie, we come to understand that Kearns invention is electronic rather than mechanical, and that it relies on simple components: a transistor, a capacitor and a variable resistor. How these work together in a novel circuit design (the Invention) is never explained in the movie. But how hard would it have been to have Kinnear explain the mechanism to his kids, using Seabrook's elegant explanation:

The resistor and the capacitor together were the timer, and the transistor worked as the switch. The resistor, which the driver could adjust with a knob, controlled the rate of current flowing into the capacitor. When the voltage in the capacitor reached a certain level, it triggered the transistor; the transistor turned on, and the wipers wiped once. The running of the wiper motor drained voltage out of the capacitor; it sank below the threshold level of the transistor, and the transistor turned off. The wipers dwelled until the capacitor recharged.

Having said all that, this is movie well worth seeing. How many times do you see circuit diagrams, even as set pieces, on the silver screen? How many times do engineers star in a film? Kinnear gives a terrific performance. And the ending, though happy enough, underscores the price the late professor Kearns and his family paid for his obsession. See the movie. Then read the article.

Digital TV preview hints at problems; firefighters come to the rescue

11.Dig.TV.Blog.gif Last month broadcasters in Wilmington, N.C., turned off their analog signals, meaning that viewers of over-the-air television had digital television or nothing. This is a preview of the nationwide analog shutdown scheduled for 17 February 2009.

Local government officials worked hard to get the word out, and an estimated 97 percent of Wilmington residents knew about the analog shut-off. The local fire department sent volunteers out to help people hook up their converter boxes. It still didnâ''t go so well. Many viewers lost their favorite television channels altogether; of the 1828 people who complained to the FCC in the first five days after the shutoff, more than half of those had lost channels. Others called their local television stations to complain.

According to a team of students from Elon University, many problems were related to the antenna. One Wilmington resident quoted in a great blog post about the antenna problems said, â''I feel scammed by all these commercials and companies. If getting a new antenna was something they knew we might have to do, why did they not say our antennas would not work?â''

So it looks like my unhappy experience in trying to switch to digital was not, unfortunately, an aberration. Just making some rough calculations, I figure that 12 percent of the roughly 15,000 people in Wilmington who donâ''t subscribe to cable or satellite were ticked off enough to call the FCC. Nationwide, twelve percent of the 13.4 million households is 1.6 million. The FCC is going to have a really busy February.

And it might not just end there. In fact, I hope those volunteer firemen keep standing by. Back in August I converted my mother in New Jersey to digital; all went well, she got lots of channels, she was happy. I just found out that, however, in spite of my carefully written instructions, a couple of days after I left she pushed some button out of sequence and hasnâ''t been able to tune in a TV signal since; wonder if I should have her call her local fire department to sort it out?

For more tales from the digital television transition, as well as links to in depth coverage about digital television technology, see IEEE Spectrum's Special Report: THE DAY ANALOG TV DIES.

Spectrum visits the Adobe Advanced Technology Labs

Clipboard-6.png

Iâ''m the family archivist, that person who takes the photos, makes the videos, sends prints to the grandparents, digs out the baby or pet pictures whenever theyâ''re needed for a school project, and, in general, is responsible for creating, storing, and organizing our familyâ''s memories.

Iâ''m reasonably successful at the creating end; I usually manage to grab a camera when Iâ''m running out to an event that ought to be commemorated, I pose the kids on birthdays and major holidays.

Iâ''m not so great at the storing and organizing part. The last time I made any serious attempt to organize photos was when I was last on maternity leaveâ''my youngest is now 10. I do regularly download digital images from my camera to my computer, occasionally back them up using an online service. I havenâ''t attempted to label any of these images since 2003, deciding it was way too time consuming. I used to order prints occasionally, but after the unopened boxes started piling up, stopped that as well.

And Iâ''ve been feeling really guilty about it; I know I really need to get my photos under control, but itâ''s just too overwhelming, so I procrastinate, and feel guilty about procrastinating.

But this week, thanks to a few engineers at Adobeâ''s Advanced Technology Lab, Iâ''m no longer feeling guilty. Oh, Iâ''m still procrastinating, but now I have an excuse.

Let me explain.

Tom Malloy, the senior vice president in charge of Advanced Technology for Adobe, invited me to visit him and a few research engineers in San Jose for a private show-and-tell. I had no idea what I might see. Malloy pulled a few projects out of the 80-some in the works and the principal engineers on the projects gave me short demos.

I expected to see interesting technology; I did, but I had a hard time keeping focused on the image processing and artificial intelligence technology that made the demos work because I got so wrapped up in what the software being demonstrated could do, namely, clean up my photo mess. Even better, it gave me an excuse to not spend days dealing with the mess myself; instead, Adobe convinced me I just have to be patient and they will make everything better. (I wondered if they are working on any tools for automating the college application process, another source of family stress these days.)

First, I saw something with the working title â''Massive Mediaâ''; it takes a userâ''s entire photo library, looks at whatever metadata is available, and sorts everything onto a timeline, grouping things by events, helping you tag large groups of photos, and restacking images for better navigation as you zoom in and out. Iâ''ve seen other attempts to stick photos on a timeline, they werenâ''t compelling; this was, mostly because of the speed at which you could navigate around the timeline and within the groups of photos. (Check out the screen shot that opens this post.)

Next up was â''Breeze,â'' a tool that uses facial recognition software to partially automate the process of tagging photos. Since, in my photo collection, Iâ''m usually searching for photos of specific family members, this would simplify my life tremendously. Again, the speed at which a large photo collection could be processed with very few clicks and keystrokes was impressive.

Finally, â''Haystackâ'' deals with pictures that have no metadata; given my first digital camera generated little automatically, thatâ''s about half of my collection. Haystack starts with some basic intelligence about scenery (beach, snow, garden, etc.); you can add more, or simply pick one photo and tell it to â''find more pictures like this.â'' Since as family archivist Iâ''ve taken a lot of shots in which the scene stays the same but the kids grow up (back to school pictures on the front porch, Halloween pictures in front of the fireplace, Christmas photos next to the tree); Iâ''d have a blast creating sequences of images. I did this once by hand (searching for back to school photos to make into a poster) and it took forever; again, Haystack seemed, at least in the demo, blazingly fast.

Inspired, I went home from the Adobe visit and added my summer photos to the 5 gigabytes of images already sitting on my computer, doing so without the usual feeling of tossing my family memories into a black hole. Thanks, Tom.

Original photography copyright 2005-2008 Heather Parker. Used with permission.

Paris Mondial de l'Automobile Plugs the Plug

Five years ago Toyota relaunched its Prius with a Saatchi & Saatchi ad blitz with the EV-bashing tagline "and you never have to plug it in." Toyota's corporate marketing manager said the idea was to show the Prius was, "not an idea that's ahead of its time."

What a difference a few years can make. At this year's Paris Mondial de l'Automobile, which opened to the press yesterday, plug-in hybrids and full-battery EVs are everywhere -- and their plugs are displayed conspicuously.

Smart, the Daimler/Swatch joint venture, towered a dangling plug over their floorspace to highlight its development of an EV model of the tiny trendy Smart Car due out in 2010. GM executives gamely held the cord of the Chevy Volt plug-in hybrid for photographers. And check out the plug on Ligier Automobiles' EV city car!

Frank Weber, GM's Global Vehicle Line Executive for the Volt, explained the shift to me in dollars and cents, or rather euros and centimes. "If you say that the charge costs less than a euro per day, it's that simple," says Weber. "Plugging in means saving, being able to drive and don't watch the signs at the gas station. This is what the plug means. It's now looked at as an opportunity and like, well ok at night you have to plug it in but you would do this anytime because the moment you plug it in you know that you save."

Americans Know Little about Nanotechnology and Less about Synthetic Biology

I am beginning to think that the Project on Emerging Technology gets some pleasure in demonstrating how ill informed US citizens are about subjects outside the scope of Britney Spearâ''s marital woes.

They have just released another poll, as they did last year, that indicates that Americans (North Americans, I presume) donâ''t know much about nanotechnology, and this year seem to know even less about synthetic biology.

This year the news reports didnâ''t come with penetrating, albeit condescending, insights such as people with less education were less likely to know about nanotechnology than those with advanced educations. But it did manage to come with foreboding tales as it did last year with the concept of â''backlashâ''.

This time around the scary scenario will be of the next presidential administration being called upon to make decisions about synthetic forms of life.

One of these days, when I have some extra cash, I am going to commission one of these polling companies to do a survey that lists every policy issue that will impact everyone and find out the degree to which people actually know the subject. I am betting that on everything from healthcare to taxes that fewer than 50% will have any idea about the subject.

Should we be alarmed? Definitely, yes. An uniformed electorate is the Achilles Heel of democracy. But I am not terribly worried that 90% of people in the US donâ''t know about synbio. However, it does get me a tad nervous that nearly 60% of Americans canâ''t name a single Supreme Court judge.

Mobile-phone newcomer shakes up Fiji

IMG_1896.JPG

I arrived in Fiji on Wednesday and was immediately greeted by an unexpected tech story. A taxi driver drove me from the airport to Suva, the capital, through hyper-green lushness, behind which farmland stretched out to the horizon. Tucked between the broad-leaved trees were one-story houses, some of them balanced on stilts to avoid flooding during the rainy seasons.

As we entered downtown Suva at 8 in the morning, dozens of teenagers and twenty-somethings thronged on the sidewalks, all massively perky and clad in bright red shirts. They were booster temporarily hired by Digicel, a Jamaican mobile phone company, to hype up the launch of the companyâ''s Fiji-wide GSM network that day. Dance music pounded throughout the central downtown area from the backs of Digicel pick-up trucks, a prelude to the all-day party the company was throwing for itself. Two Digicel minions jumped in front of our car at a stoplight and started wiping down the windshield.

Thousands of Fijians stood in line in front of the flagship Digicel store throughout the day. In Fiji, the incumbent carrier, Vodafone, has long had a virtual monopoly (another company also offers services, but it piggybacks off of Vodafoneâ''s network) and charged prohibitively high rates, according to a few locals I approached.

The network is touted as the first nation-wide service, with 95 percent coverage in Fiji. This may not be as paradigm-shattering as an iPhone launch, but I think itâ''s fair to argue that opening up the mobile phone market will have a profound impact on communications in Fiji, a collection of widespread islands that are challenging to connect to much of anything â'' a mobile phone network, a power grid, you name it. This is Digicelâ''s fifth launch in the region, following Papua New Guinea, Samoa, Tonga, and Vanuatu.

Advertisement

Tech Talk

IEEE Spectrum’s general technology blog, featuring news, analysis, and opinions about engineering, consumer electronics, and technology and society, from the editorial staff and freelance contributors.

Newsletter Sign Up

Sign up for the Tech Alert newsletter and receive ground-breaking technology and science news from IEEE Spectrum every Thursday.

Advertisement
Load More