supercomputer
Latest
Desktop-sized laser supercomputers could be coming by 2020
Small, eco-friendly optical supercomputers may soon be crunching quadrillions of calculations per second (exaflops) if a company called Optalysys has its way. It claims to be months away from demonstrating a prototype optical computer that will run at 346 gigaflops to start with -- not as fast as the best supercomputers, but pretty good for a proof-of-concept. Here's how it works: low-intensity lasers are beamed through layers of liquid crystal grids, which change the light intensity based on user inputted data. The resulting interference patterns can be used to solve mathematical equations and perform other tasks. By splitting the beam through multiple grids, the system can compute in parallel much more efficiently than standard multi-processing supercomputers (as shown in the charming Heinz Wolff-hosted video below).
IBM's Watson supercomputer will help you cook in this new recipe app
Conventional recipe apps are all well and good if you're not sure what to eat in the first place, but what if you're looking to experiment? IBM thinks its Watson supercomputer can offer some advice, so it's teaming up with the editors at Bon Appétit to test Chef Watson, an app that leans on the cognitive machine's food-making skills to spice things up. Rather than make you choose from a small, predefined set of recipes, you set some criteria and let Watson do most of the hard work; it produces 100 meal suggestions based on both the ingredients you've allowed and the cooking styles you'd prefer.
Supercomputer passes the Turing test by mimicking a teenager (update: reasons to be cautious)
After 64 long years, it looks like a machine has finally passed the Turing test for artificial intelligence. A supercomputer in a chat-based challenge fooled 33 percent of judges into thinking that it was Eugene Goostman, a fictional 13 year old boy; that's just above the commonly accepted Turing test's 30 percent threshold. Developers Vladimir Veselov and Eugene Demchenko say that the key ingredients were both a plausible personality (a teen who thinks he knows more than he does) and a dialog system adept at handling more than direct questions.
IBM's Watson computer makes a delicious BBQ sauce
We know through first-hand experience that IBM's Watson supercomputer can make a fine meal, but it's apparently an ace at condiments, too. Fast Company has tried Bengali Butternut BBQ Sauce, a recipe chosen by Watson to maximize flavor through complimentary (and fairly uncommon) ingredients. The result is a "delicious" concoction unlike what you'd normally throw on your food -- butternut and white wine give it a sweet taste, while tamarind and Thai chiles add punch that lasts beyond your last bite.
IBM's Watson supercomputer can help settle your debates
IBM's Watson supercomputer is already good at finding answers to tough questions, but it's going one step further: it can now argue an issue when there's no clear answer. A new Debater feature lets the machine take a given topic, scan for relevant articles, and automatically deduce the pros and cons based on the context and language of any claims. In a demo, Watson took 45 seconds to scour millions of Wikipedia articles and make cases both for and against limiting access to violent video games. It's likely that many people would take much longer, even if they're well-informed on the subject.
IBM's Watson supercomputer will soon be your personal shopper
Watson had been a doctor, a geneticist, a game show contestant and even a chef in the past. But now IBM's supercomputer has a new career: personal shopping. IBM has partnered with digital commerce firm Fluid to develop a cloud-based app called Expert Personal Shopper (XPS), which uses Watson's brains to answer buyers' highly specific questions. In short, the computer with many hats now plays the role of a sales associate when you're shopping online. IBM and Fluid are currently working with several consumer brands, but The North Face will be the first to feature the technology on its website. When the outdoor clothing and equipment company launches XPS, you can ask it questions like you would an assistant at a mall. If you needed a recommendation on the best equipment to use for a five-day cross-country trip, or need to know the best tent to use if you're hiking with family, including kids, then Watson's got your back. It's unclear when XPS will launch exactly, but IBM has allotted $100 million in funds to develop that and a variety of other cognitive apps. All parties involved are planning to develop it further for mobile applications and devices.
NVIDIA announces the Jetson TK1 dev-kit, calls it the world's first mobile supercomputer
Wish you had your own personal supercomputer? Soon, you'll be able to buy one -- well, sort of. At its GPU Technology conference today, NVIDIA announced the Jetson TK1, a $192 Tegra K1-based development kit built on the same architecture that powers the Titan supercomputer at Oak Ridge National Laboratories. NVIDIA CEO Jen-Hsun Haung describes it as "the world's tiniest little supercomputer," noting that it's capable of running anything the Titan can run, but at a much slower pace.
IBM sends Watson on a genetic quest to find the best cancer treatments
Sure, IBM's Watson crunches data for mobile apps and powers food trucks, but its owners are constantly looking for important studies that can put its cognitive computing expertise to the test. With the recent announcement of a clinical trial studying ways to deliver personalized care to brain cancer patients, the Jeopardy-conquering supercomputer appears to have found that next major challenge. In collaboration with New York Genome Center, Watson will be tasked with trawling archives of medical literature and clinical data, using its patten recognition skills to identify the best cancer treatments based on a patient's genetic make-up. Teams of scientists had manually undertaken the process before, but it's exactly the kind of problem Watson was designed to help solve. IBM says it will begin a trial later this year and hopes to open its findings to doctors across the world.
Google tests the performance limits of D-Wave's quantum computers
We've long known that D-Wave's quantum computers are specialized tools rather than Swiss Army Knives, but just how good are they at their intended tasks? Google has just conducted some benchmarking to find out, and the short answer is that these systems are very good -- but they have definite limits. A current-generation D-Wave 2 is about 35,500 times faster than a generalized problem-solving computer when both are running standard software. However, some of that advantage disappears when a general-purpose computer runs code that simulates quantum computing. While D-Wave's hardware is better at dealing with structured code, it runs neck-and-neck with the "fake" system when tackling random problems. Not that Google is feeling much in the way of buyer's remorse. It believes that further tests could see the D-Wave unit come out ahead, and future quantum machines should make it harder for conventional PCs to catch up.
NVIDIA unveils Tesla K40 accelerator, teams with IBM on GPU-based supercomputing
NVIDIA's Tesla GPUs are already mainstays in supercomputers that need specialized processing power, and they're becoming even more important now that the company is launching its first Tesla built for large-scale projects. The new K40 accelerator only has 192 more processing cores than its K20x ancestor (2,880, like the GeForce GTX 780 Ti), but it crunches analytics and science numbers up to 40 percent faster. A jump to 12GB of RAM, meanwhile, helps it handle data sets that are twice as big as before. The K40 is already available in servers from NVIDIA's partners, and the University of Texas at Austin plans to use it in Maverick, a remote visualization supercomputer that should be up and running by January. As part of the K40 rollout, NVIDIA has also revealed a partnership with IBM that should bring GPU-boosted supercomputing to enterprise-grade data centers. The two plan on bringing Tesla GPU support to IBM's Power8-based servers, including both apps and development tools. It's not clear when the deal will bear fruit, but don't be surprised if it turbocharges a corporate mainframe near you.
IBM preparing to launch a Watson cloud service, lease out APIs to developers
Need a little help from a supercomputer? You just might get it: IBM is getting ready to release a more powerful version of its Watson supercomputer, and it'll be available for rent. According to the PC World, the company is working on a Watson-based cloud service that developers can use to build richer, more interactive applications. The project uses a smaller, more scaleable version of Watson to build cognitive supercomputers as needed -- potentially leaving a smaller footprint. IBM is already working with partners to kick the service off, specifically Fluid, who is creating a Watson-powered retail assistant that's said to actively converse with customers to help them make "more informed buying decisions." Developers familiar with RESTful APIs shouldn't have too much trouble using Watson, IBM's Rob High told the PC World, although it won't be exactly like traditional programming. "Cognitive systems are different in that they have the ability to simulate human behavior. For the most part humans have had to adapt to the computer. As we get into cognitive systems we open up the aperture to the computer adapting to the human." IBM has yet to specifics on when the Watson-powered cloud will be available, or how much it'll cost developers to work with the Jeopardy star. Hopefully, the cognitive computer will still be able to put together a tasty pastry.
Parallella 'supercomputers' headed to early backers, 16-core boards up for general pre-order
Following its successful Kickstarter campaign, Adapteva flashed the production versions of its Parallella "supercomputer" boards in April, penning in a loose summer delivery date. Today, the company reports that the first "beta" units have begun winding their way to backers who pledged at the DEVELOPER, 64-CORE-PLUS and ROLF levels. Other backers should receive their boards by summer's end "after some final refinements." For those who missed the crowd-funding window, you too can get a Parallella, as Adapteva has now opened up general pre-orders for the 16-core version on its website. While all Kickstarter-bought boards will bear a Zynq-7020 SoC, new pre-orders are configured with a 7010 as standard, though you can upgrade to the 7020 if you lay down a little more dough. However, newcomers will be treated to "Gen-1" boards, which will offer slight improvements over earlier versions, such as reduced power consumption and an added serial port three-pin header. You'll find the basic 16-core board going for $99 over at Adapteva's store, with an expected October delivery date. The company tells us the 64-core version will also be available for public consumption, with pre-orders beginning in Q4 this year.
USC finds that D-Wave's quantum computer is real, maybe
D-Wave has had little trouble lining up customers for its quantum computer, but questions have persisted as to whether or not the machine is performing quantum math in the first place. University of Southern California researchers have tested Lockheed Martin's unit to help settle that debate, and they believe that D-Wave's computer could be the real deal -- or rather, that it isn't obviously cheating. They've shown that the system isn't based on simulated annealing, which relies on traditional physics for number crunching. The device is at least "consistent" with true quantum annealing, although there's no proof that this is what's going on; it may be using other shortcuts. Whether or not D-Wave built a full-fledged quantum computer, the resulting output is credible enough that customers won't feel much in the way of buyer's remorse.
Tianhe-2 supercomputer claims the lead in Top 500 list, thanks its 3.1 million processor cores
As predicted, Chinese supercomputer Tianhe-2 (also known as the Milky Way-2) has now been crowned the most powerful supercomputer in the world. Arriving years ahead of schedule, and packing 32,000 Xeon processors alongside 48,000 Xeon Phi accelerator processors, the supercomputer can manage a quadrillion mathematical calculations per second (33.85 petaflops), double that of last year's king (and closest rival), the Titan. In this year's results, 80 percent of the Top 500 used Intel processors, while 67 percent had processors with eight or more cores -- as clock speeds stall, supercomputer development has now focused on processors running in parallel. Top 500 editor Jack Dongarra adds that "most of the features of the [Tianhe-2] system were developed in China, and they are only using Intel for the main compute part," meaning that you can expect to see more Chinese entrants (and possibly champions) over the next few years. For now, however, the US still claims the majority of the Top 500, with 253 top-ranking supercomputers.
Tianhe-2 may easily crush supercomputer speed record at 30.7 petaflops
Many suspect that China's Tianhe-2 could win the supercomputer speed wars, but there haven't been real numbers to back up that hunch. We now have some of those figures courtesy of Top 500's Jack Dongarra, and Tianhe-2 could well be the new leader -- by a gigantic margin. The cluster of Ivy Bridge and Xeon Phi chips has benchmarked at 30.65 petaflops when using 90 percent of its nodes, giving it a 74 percent edge (!) over the 17.6-petaflop Titan. There's no guarantee that Tianhe-2 will hold the crown when the official Top 500 rankings appear on June 17th, but we don't see any upstart rivals on the horizon. It could be lonely at the top... for a while.
Sequoia supercomputer breaks simulation speed record, 41 times over
While we've seen supercomputers break records before, rarely have we seen the barrier smashed quite so thoroughly as by Lawrence Livermore National Laboratory's Sequoia supercomputer. Researchers at both LLNL and Rensselaer Polytechnic Institute have used planet-scale calculations on the Blue Gene/Q-based cluster to set an all-time simulation speed record of 504 billion events per second -- a staggering 41 times better than the 2009 record of 12.2 billion. The partnership also set a record for parallelism, too, by making the supercomputer's 1.97 million cores juggle 7.86 million tasks at once. If there's a catch to that blistering performance, it's not knowing if Sequoia reached its full potential. LLNL and RPI conducted their speed run during an integration phase, when Sequoia could be used for public experiments; now that it's running classified nuclear simulations, we can only guess at what's possible.
Titan supercomputer to be loaded with 'world's fastest' storage system
If you figured Titan's title of the world's most powerful supercomputer would give the folks at Oak Ridge National Laboratory reason to rest on their laurels, you'd be mistaken. The computer is set to have its fleet of 18,688 NVIDIA K20 GPUs and equal number of AMD Opteron processors paired with what's said to be the planet's speediest storage system, making its file setup six times faster and giving it three times more capacity. Dubbed Spider II, the new hardware will endow the number cruncher with a peak performance of 1.4 terabytes a second and 40 petabytes of storage spread across 20,000 disk drives. Behind the refresh are 36 of Datadirect Networks' SFA12K-40 systems, which each pack 1.12PB of capacity. For more on the herculean rig's upgrade, hit the jump for the press release.
IBM Roadrunner retires from the supercomputer race
For all the money and effort poured into supercomputers, their lifespans can be brutally short. See IBM's Roadrunner as a textbook example: the 116,640-core cluster was smashing records just five years ago, and yet it's already considered so behind the times that Los Alamos National Laboratory is taking it out of action today. Don't mourn too much for the one-time legend, however. The blend of Opteron and Cell processors proved instrumental to understanding energy flow in weapons while also advancing the studies of HIV, nanowires and the known universe. Roadrunner should even be useful in its last gasps, as researchers will have a month to experiment with the system's data routing and OS memory compression before it's dismantled in earnest. It's true that the supercomputer has been eclipsed by cheaper, faster or greener competitors, including its reborn Cray arch-nemesis -- but there's no question that we'll have learned from Roadrunner's brief moment in the spotlight.
University of Illinois' Blue Waters supercomputer now running around the clock
Things got a tad hairy for the University of Illinois at Urbana-Champaign's Blue Waters supercomputer when IBM halted work on it in 2011, but with funding from the National Science Foundation, the one-petaflop system is now crunching numbers 24/7. The behemoth resides within the National Center for Supercomputing Applications (NCSA) and is composed of 237 Cray XE6 cabinets and 32 of the XK7 variety. NVIDIA GK110 Kepler GPU accelerators line the inside of the machine and are flanked by 22,640 compute nodes, which each pack two AMD 6276 Interlagos processors clocked at 2.3 GHz or higher. At its peak performance, the rig can churn out 11.61 quadrillion calculations per second. According to the NCSA, all that horsepower earns Blue Waters the title of the most powerful supercomputer on a university campus. Now that it's cranking away around-the-clock, it'll be used in projects investigating everything from how viruses infect cells to weather predictions.
Watson ponders careers in cooking, drug research as IBM makes it earn its keep
While mad game show skills are nice and all, IBM has started to nudge Watson toward the door to begin paying its own freight. After a recent foray into finance, the publicity-loving supercomputer has now brought its number-crunching prowess to the pharmaceutical and pastry industries, according to the New York Times. If the latter sounds like a stretch for a hunk of silicon, it actually isn't: researchers trained Watson with food chemistry data, flavor popularity studies and 20,000 recipes -- all of which will culminate in a tasting of the bot's freshly devised "Spanish Crescent" recipe. Watson was also put to work at GlaxoSmithKline, where it came up with 15 potential compounds as possible anti-malarial drugs after being fed all known literature and data on the disease. So far, Watson projects haven't made Big Blue much cash, but the company hopes that similar AI ventures might see its prodigal child finally pay back all those years of training.
