supercomputer
Latest
NSA wants $896.5 million to build new supercomputing complex
The federal government may be cutting corners left and right, but that hasn't stopped the NSA from requesting nearly $900 million to help beef up its supercomputing capabilities. According to budget documents released by the Department of Defense yesterday, the NSA is looking to construct a massive new High Performance Computing Center in Maryland, designed to harness plenty of supercomputing muscle within an energy efficient framework. As with many other data centers, the NSA's $896.5 million complex would feature raised floors, chilled water systems and advanced alarm mechanisms, but it would also need about 60 megawatts of power -- the same amount that powers Microsoft's gargantuan, 700,000 square-foot data center in Chicago. According to the DoD, however, the NSA would use that juice judiciously, in the hopes of conserving enough water, energy and building materials to obtain LEED Silver certification. Another chunk of the funding, not surprisingly, would go toward fortifying the facility. The NSA is hoping to pour more than $35 million into building security and perimeter control, which would include a cargo inspection facility, advanced surveillance, and systems designed to detect any radiological, nuclear, or chemical threats. If all goes to plan, construction would wrap up by December 2015.
IBM touts new Power7 systems, still no mass market Watson
IBM's latest announcement probably won't get us any closer to securing our very own in-house version of Watson, but the firm is boasting a new line of Power7 products that includes an upgraded version of the supercomputer's server. First up are the BladeCenter PS703 and PS704, sporting 16 cores and 32 cores, respectively -- the PS704 touts a 60 percent increase in speed over its predecessors. The Power 750, the same system that gave Watson the stuff to slaughter those humans on Jeopardy!, is getting an upgrade that supports as many as 32 cores and can run up to 128 simultaneous threads, while the Power 755 offers up high-performance computing with 32 cores of its own. The cheapest version of the Power 750 Express rings in at about $30,000. So, no, we won't be battling Watson in a Jeopardy! Home Edition showdown anytime soon, but we're happy to see that our favorite supercomputer could be even smarter -- or at the very least, faster -- the next time it shows up on the boob tube. Full PR after the break.
Ken Jennings talks about losing to Watson, being human after all
In a piece for Slate titled "My Puny Human Brain," former-Jeopardy-greatest Ken Jennings talks briefly through his experience playing against IBM's Watson. If you were hoping for some sour grapes, you won't find it here, but Ken gives a great insight into what it feels like to be an underdog human up against a PR darling supercomputer. "Watson has lots in common with a top-ranked human Jeopardy! player: It's very smart, very fast, speaks in an uneven monotone, and has never known the touch of a woman." Ken wraps it up on an uplifting, humans-are-going-to-be-alright-after-all note, and we seem to have something in our eye...
Conan kills Watson, evening the score (video)
Sure, IBM's Watson is able to beat humans to the buzzer on some lousy television game show. But how does the smarmy supercomputer hold up to the business end of a baseball bat wielded by a jealous husband with opposable thumbs? Click through the break for a taste of sweet, carbon-based revenge.
Watson wins it all, humans still can do some other cool things
In case you missed it, Watson won again tonight. He even got the Final Jeopardy question correct this time, a multi-layered reference to Bram Stroker that he bet $10k on. His final score over the two rounds ended up at $77,147 (Watson has this thing for betting strange amounts that usually end in a 7), while Ken Jennings got $24,000 and Brad Rutter did $21,600 -- both humans saving a bit of face after last round's stunning defeat. Watson will be giving his $1,000,000 winnings to charity. So, a few things: We're totally surprised, in a larger theoretical sense, that a computer could win at Jeopardy. We're totally not surprised that Watson, the system built by IBM over the past few years at the expense of millions of dollars, actually succeeded at winning at Jeopardy. Computers have better reflexes than humans, as it turns out. Deal with it. If you can't tell, we're having a little trouble processing all the emotions brought on by a Jeopardy win from IBM's Watson supercomputer. It's obvious that IBM's DeepQA research program has developed some of the most sophisticated natural language AI known to man. At the same time, Jeopardy questions aren't really that hard. As evidenced by watching these Watson-dominated matches, all three contestants knew the answer most of the time, but Watson was just quicker on the draw. Of course, it's no surprise that computers have quicker reflexes (even with the "handicap" of having to mechanically press the same style of clicker as Meatbag 001 and Meatbag 002), so why shouldn't Watson get to use his inbuilt advantage to the utmost? It seems like a fair fight to us. The question of "who is better at Jeopardy" aside (trust us, it's Watson), the larger implications for the human race and our computer sidekicks are still unclear. Watson can currently answer simple trivia questions, sometimes couched in puns or minor riddles, with a decent level of accuracy. The answers themselves are no more than a high school student with Wikipedia access could pull off, and Watson has no way of knowing for sure when he's right. He lacks a solid, computer-readable database of "facts" like a Wolfram Alpha, or the incredible reasoning abilities of a human, instead relying on statistical analysis of vast amounts of text. When it comes to Jeopardy, it turns out to be Good Enough, which is actually a pretty incredible achievement in the world of AI, and we're sure we'll be finding out soon what other applications IBM thinks Watson is Good Enough at -- they're thinking everything from healthcare to the financial industry. Still, we're sure some of us clicker-speed-nit-pickers will remain unimpressed. Make sure to check out the Engadget Show tomorrow, where we'll be chatting up the creators of Watson about all this, but for now... 01000011 01101111 01101110 01100111 01110010 01100001 01110100 01110011 00100001
Humans had a good run: Watson to debut on Jeopardy tonight
Today, tomorrow, and the 16th are the fateful days: IBM's Watson supercomputer will go head to head with Brad Rutter and Ken Jennings in this silly, human-devised game we call Jeopardy. It promises to be some kickass TV, at the very least, and a historic event if Watson can prevail over his fleshy competition. The two matches, which are being spread over the three days, were pre-taped, so Ken, Brad, Alex and Watson already know the outcome, but they've done a pretty good job of keeping the secret so far. Hopefully they can keep mum until 7pm-ish this evening (check your local listings for a specific time). Need something to keep you occupied until then? Check out the great Esquire feature on Ken Jennings at the More Coverage link below. Oh, and don't miss our coverage of last month's preview match.
IBM's Mira supercomputer does ten petaflops with ease, inches us closer to exascale-class computing
Say hello to the Blue Gene/Q, or if you're looking for something a bit less intimidating, "Mira." That's IBM's latest and greatest concoction, a ten-petaflop supercomputer capable of running programs at ten quadrillion calculations a second. Hard to say who'd win between Mira and Watson, of course, but there's absolutely no question who'd come out on top if Mira were pitted against her predecessor Intrepid (hint: Mira's 20x faster). To put this all in perspective, IBM's chiming in with this: "If every man, woman and child in the United States performed one calculation each second, it would take them almost a year to do as many calculations as Mira will do in one second." Mira's next stop is at the US Department of Energy's Argonne National Laboratory, where it'll be used to tackle 16 projects in particular that were drawn from a pool of proposals to gain access to her capabilities. We're told that these include a range of initiatives -- from reducing energy inefficiencies in transportation and developing advanced engine designs to spurring advances in energy technologies -- and in time, it could lead to exascale-class computers "that will be faster than petascale-class computers by a factor of a thousand." And here we are getting excited about a 5GHz Core i7.
IBM's Watson supercomputer will play Jeopardy! on these dates
What are February 14th, 15th, and 16th? We've known it was going to happen for a while, but now we know when to set our DVRs. A rack of servers -- soaked with natural-language processing, armed with a battalion of esoteric pop culture knowledge, and "represented by a round avatar" -- will face off against Jeopardy! millionaires Ken Jennings and Brad Rutter (both profiled in video after the break) for three days starting on Valentine's Day 2011. We're also hearing that Watson will sign autographs after it's done decimating its opponents. It will then donate all its winnings to charity and spend the rest of its natural life dodging paparazzi on an undisclosed beach in the South Pacific. [Photo from Ben Sisto's flickr]
IBM breakthrough brings us one step closer to exascale computing, even more intense chess opponents
The path to exascale computing is a long and windy one, and it's dangerously close to slipping into our shunned bucket of "awesome things that'll never happen." But we'll hand it to IBM -- those guys and gals are working to create a smarter planet, and against our better judgment, we actually think they're onto something here. Scientists at the outfit recently revealed "a new chip technology that integrates electrical and optical devices on the same piece of silicon, enabling computer chips to communicate using pulses of light (instead of electrical signals), resulting in smaller, faster and more power-efficient chips than is possible with conventional technologies." The new tech is labeled CMOS Integrated Silicon Nanophotonics, and if executed properly, it could lead to exaflop-level computing, or computers that could handle one million trillion calculations per second. In other words, your average exascale computer would operate around one thousand times faster than the fastest machine today, and would almost certainly give Garry Kasparov all he could stand. When asked to comment on the advancement, Dr. Yurii A. Vlasov, Manager of the Silicon Nanophotonics Department at IBM Research, nodded and uttered the following quip: "I'm am IBMer, and exascale tomfoolery is what I'm working on."* *Not really, but you believed it, didn't you?
China's Tianhe-1A is world's fastest supercomputer, plans to usurp the West now complete
It happened. China just passed the US and the world with the reveal of the world's fastest supercomputer. The fully operational Tianhe-1A, located at the National Supercomputer Center in Tianjin, scored 2.507 petaflops as measured by the LINPACK benchmark. That moves it past Cray's 2.3 petaflops Jaguar located at Oak Ridge National Lab in Tennessee. Tianhe-1A achieved the record using 7,168 NVIDIA Tesla M2050 GPUs and 14,336 Intel Xeon CPUs consuming 4.04 megawatts. Knowing that 10 petaflops is within reach by 2012, we'll see if Tianhe-1A can maintain its title when the new Top500 supercomputers list is released next week.
Fujitsu K supercomputer will do 10 petaflops in 2012, eat Crays for breakfast
10's a nice round number, isn't it? Round, yes, but also wildly impressive when you put the word "petaflops" behind it as Fujitsu has done with its upcoming K supercomputer, which will be able to crunch through 10 quadrillion operations every second. Compare that to the current champ of processing farms, Cray's Jaguar, which can handle only (only!) 1.75 petaflops of workload and you'll know that we're talking about a seminal leap in performance. Japan's Riken Research Institute is the fortunate addressee on the crates of ultrafast SPARC64 VIIIfx processors that Fujitsu is now shipping out and the current plan is to have everything up and running by 2012. In total, there'll be 80,000 CPUs, each possessing 8 cores running at 2.2GHz, which will be housed within 800 racks. So yes, there'll be a machine somewhere on the Japanese isle with 640,000 processing cores at its disposal. Feeling safe?
MIT app turns your Android phone into a supercomputer... of sorts
Oh, sure -- a few people have called Google's Nexus One a "superphone," but suddenly, that nickname has taken on a whole new level of meaning. A team of talent from MIT has put its head down in order to concoct a new Android application that can come darn close to solving complex computational problems in just a fraction of the time that it'd take a bona fide supercomputer. The goal here is to let researchers and scientists convert to Google's mobile OS, but if you aren't falling for that one, it's also designed to "let engineers perform complicated calculations in the field, and to better control systems for vehicles or robotic systems." Of course, the models that are hosted on the phone do require a supercomputer to create, but once certain formulas are embedded, the app can then compute approximations in mere seconds rather than hours. Best of all, rbAPPmit is available for download as well speak in the source link below, but we'd probably wait for the (presumably thick) user guide to surface before diving in headfirst. [Thanks, Alasdair]
Inhabitat's Week in Green: solar robots that fly, CO2 fabric dye, and the dark silicon that boosts battery life
Each week our friends at Inhabitat recap the week's most interesting green developments and clean tech news for us -- it's the Week in Green. Solar power blazed a trail this week as we took a look at several hot new technologies, starting with SkyFuel's SkyTrough, which is being billed as the world's most efficient solar collector. We also saw solar energy take to the skies as designers unveiled plans for a fleet of high-flying solar robots, and we were surprised to learn that common household dyes could significantly increase the efficiency of photovoltaic panels by optimizing their color absorption spectrum. Speaking of dye, from the realm of wearable tech we also brought you a breakthrough new technique for dying fabric that saves water by utilizing fluid CO2. We also saw a prototype for a wired "safe cuddling" suit for kids that wards off improper touching by sounding an alarm, and if you're a fan of high-tech footwear, check out these tricked-out kicks that do double duty as Wii controllers. This week also saw a tremendous green boost for bits and bytes as the University of Leicester switched on its hyper-efficient ALICE supercomputer, which is ten times more powerful than its predecessor and stands to reduce yearly CO2 emissions by 800 tons. Meanwhile, researchers at UC San Diego revealed work on a new mobile phone chip that harnesses "dark silicon" to boost smartphone battery life by a factor of eleven. We also showcased several efficient autos as Southern California rolled out a fleet of all electric buses that can recharge in 10 minutes flat, and the hyper-miling Avion car embarked on a trip from Canada to Mexico with just 14 gallons of gas. And if you'll be doing some traveling of your own back to school this fall, you wont want to miss this chance to pick up an awesome solar-powered Sakku satchel. Finally, with Labor Day on its way why not upgrade your BBQ with an adorable altoids tin mini grill - it's curiously awesome!
Homebrew Cray-1A emulates the iconic supercomputer, to no useful purpose
The Cray-1A first made the scene in 1976, weighing 5.5 tons (including the refrigeration system) and running at 80MHz -- with a whopping 8MB RAM. Who wouldn't want to own one -- or a miniature version of one, for that matter? Chris Fenton would, apparently. Yes, it's that Chris Fenton -- the electrical engineer who once made a $50 laptop out of a PICAXE 18X Microcontroller and 96 bytes of RAM (and some wood). And he's back with a 1/10-scale Cray-1A. And unlike a similar project we've seen in the past, this bad boy runs a custom Cray emulator (too bad there doesn't seem to be any Cray software floating around). Wild, huh? Get the whole scoop (and some pointers if you want to roll your own) after the break.
DARPA enlists NVIDIA to build exascale supercomputer that's '1000x faster' than today's quickest
At this point, it's pretty obvious that GPUs will soon be playing a huge role in modern day supercomputers -- a role that may just rival that of the tried-and-true CPU. Virginia Tech is gleefully accepting $2 million in order to build a GPU and CPU-enabled HokieSpeed supercomputer, and today DARPA is handing out $25 million to NVIDIA in order to develop "high-performance GPU computing systems." Specifically the Defense Department's research and development arm is aiming to address a so-called "crisis in computing," and if all goes well, the four-year project will eventually yield a "new class of exascale supercomputers which will be 1,000-times more powerful than today's fastest supercomputers." That's a pretty lofty goal, but NVIDIA will be aided by Cray, Oak Ridge National Laboratory and a half-dozen US universities along the way. And yeah, if ever anyone's ego was prepared to topple Moore's Law, it'd be this guy.
Virginia Tech's HokieSpeed supercomputer to rely on CPU and GPU synergies
Virginia Tech's no stranger to housing supercomputers -- those folks strung together 324 Mac Pros back in 2008 just for kicks, giggles and "research" -- but their latest computing monolith is quite the shift from the ordinary. A cool $2 million is floating over to Blacksburg in order to create HokieSpeed, a "versatile new supercomputing instrument" that'll soon be primed and ready to handle not just one or two tasks, but a variety of disciplines. Wu Feng, associate professor of computer science at the university, calls this magnificent monster a "new heterogeneous supercomputing instrument based on a combination of central processing units (CPUs) and graphical processing units (GPUs)," with expected performance to be orders of magnitude higher than their previous claim to fame, System X. One of its first assignments? To give end users the ability "to perform in-situ visualization for rapid visual information synthesis and analysis," and during the late hours, hosts a campus-wide Quake deathmatch. Just kidding on that last bit... maybe.
Honey, Daryl Brach shrunk the Cray-1 supercomputer
The original Cray supercomputer, the Cray-1, is an iconic piece of computing history, so big it had a ring of padded seats around which engineers could sit and contemplate esoteric questions of life whilst the machine humming behind them answered the more finite ones. This semi-hexadecagon shape has been brought back to life, scaled down quite a bit, by case modder and woodcrafter Daryl Brach. The original 5.5 ton behemoth is now a desktop-friendly size, and though those seats are now too small for human behinds they're still leather-covered and padded, hiding a pair of DVD-ROM drives connected to not one but two motherboards. We're not sure what other hardware Brach populated the thing with internally, but given that original Cray-1 had 8MB of memory to work with we're guessing this modern version would have no problem computationally wiping the floor with its inspiration.
Inhabitat's Week in Green: of mirror cubes and urban icebergs
Each week our friends at Inhabitat recap the week's most interesting green developments and clean tech news for us -- it's the Week in Green. With summer in full swing, this week Inhabitat watched the mercury rise as the world's largest thermostat burst forth with an array of 72,000 building-mounted LEDs. We also kept things cool with a remarkable plan to transform frozen construction sites into event-hosting urban icebergs. And if you haven't made plans for a summer vacation yet, might we recommend this stunning Swedish "Treehotel" housed within a silvery mirror cube in the sky? Heartening news rang forth from the renewable energy sector this week as a UN-backed study reported that the building of new renewable energy plants has officially overtaken fossil fuel plants in Europe and the US. We also took a look at two brand new types of power plants -- the world's first hydrogen-driven power plant in Italy and the first hybrid coal-solar power plant in Colorado. The past week also saw several remarkable advances in clean tech, starting with MIT's latest innovation, a new type of high-tech fiber that can transmit sound, light, and generate electricity. We also paid homage to one of our all-time favorite sources of (surprisingly green) home entertainment - the Roku Box.
Tokyo University's Grape-DR supercomputer is a tangled green powerhouse
We live in an ecologically minded era, where Ford spends more time talking up the new Mustang's mpg rating than its 0 - 60 times. Appropriate, then, that supercomputers are now being rated not on ultimate speed but on speed relative to power consumption. Top of the Green500 supercomputer list is the Grape-DR, a Japanese cluster at the University of Tokyo powered by a combination of 128 Intel Core i7-920 processors and four bespoke accelerator chips. That combination enables the system to manage 815.43 megaflops per watt, a good bit higher than the 773.38 rating an IBM-based machine in Germany managed. That's quite a bit lower than the team hopes to achieve, indicating they can boost that rating by 50 percent by the end of the year. Hopefully by then they invest in some cable management. Two of our staff network engineers passed out after just glancing at the picture above. The third... well, he didn't fare so well.
Tokyo Institute of Technology announces SSD-packing, 2.39 petaflop supercomputer
IBM has announced plans to start using SandForce SSDs in its enterprise machines, and now it looks like the Tokyo Institute of Technology is doing one better, working with NEC and HP to produce Tsubame 2.0. This next-gen supercomputer will reportedly operate at 2.39 petaflops (that's a lot of flops!) and uses a new multilevel storage architecture consisting of DRAM as well as SSDs. Not only will this bad boy have thirty times the computing capacity of Tsubame 1.0 (due in part to its some 2,816 Intel Westmere microprocessors and 4,224 NVIDIA Tesla M2050 GPUs), its power draw should be some 1/25th of its predecessor's. If all goes according to plan, it should be in operation this fall, at a cost of ¥3.2 billion (approx $35.5 million). [Thanks, Dylan]
