Four years ago, Google was faced with a conundrum: if all its users hit its voice recognition services for three minutes a day, the company would need to double the number of data centers just to handle all of the requests to the machine learning system powering those services, reads a PCWorld article, which talks about how Tensor Processing Unit (TPU), a chip that is designed to accelerate the inference stage of deep neural networks came into being. The article shares an update: Google published a paper on Wednesday laying out the performance gains the company saw over comparable CPUs and GPUs, both in terms of raw power and the performance per watt of power consumed. A TPU was on average 15 to 30 times faster at the machine learning inference tasks tested than a comparable server-class Intel Haswell CPU or Nvidia K80 GPU. Importantly, the performance per watt of the TPU was 25 to 80 times better than what Google found with the CPU and GPU.

schwit1 quotes a report from IEEE Spectrum: The IBM 1460, which went on sale in 1963, was an upgrade of the 1401 [which was one of the first transistorized computers ever sold commercially]. Twice as fast, with a 6-microsecond cycle time, it came with a high-speed 1403 Model 3 line printer. The 1403 printer was incredibly fast. It had five identical sets of 48 embossed metal characters like the kind you'd find on a typewriter, all connected together on a horizontal chain loop that revolved at 5.2 meters per second behind the face of a continuous ream of paper. Between the paper and the character chain was a strip of ink tape, again just like a typewriter's. But rather than pressing the character to the paper through the ink tape, the 1403 did it backward, pressing the paper against the high-speed character chain through the ink tape with the aid of tiny hammers. Over the years, IBM came out with eight models of the 1403. Some versions had 132 hammers, one for each printable column, and each was individually actuated with an electromagnet. When a character on the character chain aligned with a column that was supposed to contain that character, the electromagnetic hammer for that column would actuate, pounding the paper through the ink tape and into the character in 11 microseconds. With all 132 hammers actuating and the chain blasting along, the 1403 was stupendously noisy [...] The Model 3, which replaced the character chain with slugs sliding in a track driven by gears, took just 55 milliseconds to print a single line. When printing a subset of characters, its speed rose from 1,100 lines per minute to 1,400 lines per minute.

An anonymous reader quotes a report from The Verge: Norway's famous doomsday seed vault is getting a new neighbor. It's called the Arctic World Archive, and it aims to do for data what the Svalbard Global Seed Vault has done for crop samples -- provide a remote, impregnable home in the Arctic permafrost, safe from threats like natural disaster and global conflicts. But while the Global Seed Vault is (partially) funded by charities who want to preserve global crop diversity, the World Archive is a for-profit business, created by Norwegian tech company Piql and Norway's state mining company SNSK. The Archive was opened on March 27th this year, with the first customers -- the governments of Brazil, Mexico, and Norway -- depositing copies of various historical documents in the vault. Data is stored in the World Archive on optical film specially developed for the task by Piql. (And, yes, the company name is a pun on the word pickle, as in preserving-in-vinegar.) The company started life in 2002 making video formats that bridged analog film and digital media, but as the world went fully digital it adapted its technology for the task of long-term storage. As Piql founder Rune Bjerkestrand tells The Verge: "Film is an optical medium, so what we do is, we take files of any kind of data -- documents, PDFs, JPGs, TIFFs -- and we convert that into big, high-density QR codes. Our QR codes are massive, and very high resolution; we use greyscale to get more data into every code. And in this way we convert a visual storage medium, film, into a digital one." Once data is imprinted on film, the reels are stored in a converted mineshaft in the Arctic archipelago of Svalbard. The mineshaft (different to the one used by the Global Seed Vault) was originally operated by SNSK for the mining of coal, but was abandoned in 1995. The vault is 300 meters below the ground and impervious to both nuclear attacks and EMPs. Piql claims its proprietary film format will store data safely for at least 500 years, and maybe as long as 1,000 years, with the assistance of the mine's climate.

An anonymous reader quotes a report from Bloomberg: General Motors has connected about a quarter of its 30,000 factory robots to the internet, and the largest U.S. automaker already is reaping the benefits of less down time. In the last two years, GM has avoided 100 potential failures of vehicle-assembling robots by analyzing data they sent to external servers in the cloud, Mark Franks, director of global automation, said at a conference in Chicago on Monday. Connectivity is preventing assembly line interruptions and robot replacements that can take as long as eight hours. Internet monitoring allows GM to order parts when it detects they're wearing out instead of having to store them at the factory. That reduces inventory and saves money, Franks said. Hooking robots to the internet for preventive maintenance is just the start of a spurt of new robotics technology, Franks said. GM is using robots that can work safely alongside humans in the factory that produces the Chevrolet Volt plug-in hybrid, he said.

An anonymous reader writes from a report via BleepingComputer: Last week, at the BlackHat Asia 2017 security conference, researchers from cyber-security firm Cylance disclosed two vulnerabilities in the firmware of Gigabyte BRIX small computing devices, which allow an attacker to write malicious content to the UEFI firmware. During their presentation, researchers installed a proof-of-concept UEFI ransomware, preventing the BRIX devices from booting, but researchers say the same flaws can be used to plant rootkits that allow attackers to persist malware for years. The two vulnerabilities discovered are CVE-2017-3197 and CVE-2017-3198. The first is a failure on Gigabyte's part to implement write protection for its UEFI firmware. The second vulnerability is another lapse on Gigabyte's side, who forgot to implement a system that cryptographically signs UEFI firmware files. Add to this the fact that Gigabyte uses an insecure firmware update process, which doesn't check the validity of downloaded files using a checksum and uses HTTP instead of HTTPS. A CERT vulnerability note was published to warn users of the impending danger and the bugs' ease of exploitation.

Samsung has been working on its Tizen operating system for several years now, implementing it into its various televisions and smartwatches. According to a report from Motherboard, the OS isn't receiving a lot of praise in the security department. Israeli researcher Amihai Neiderman has found 40 unknown zero-day vulnerabilities in Tizen, adding that it may be the worst code he's ever seen. From the report: "It may be the worst code I've ever seen," he told Motherboard in advance of a talk about his research that he is scheduled to deliver at Kaspersky Lab's Security Analyst Summit on the island of St. Maarten on Monday. "Everything you can do wrong there, they do it. You can see that nobody with any understanding of security looked at this code or wrote it. It's like taking an undergraduate and letting him program your software." All of the vulnerabilities would allow hackers to take control of a Samsung device from afar, in what's called remote-code execution. But one security hole Neiderman uncovered was particularly critical. It involves Samsung's TizenStore app -- Samsung's version of Google Play Store -- which delivers apps and software updates to Tizen devices. Neiderman says a flaw in its design allowed him to hijack the software to deliver malicious code to his Samsung TV. Because the TizenStore software operates with the highest privileges you can get on a device, it's the Holy Grail for a hacker who can abuse it. Although TizenStore does use authentication to make sure only authorized Samsung software gets installed on a device, Neiderman found a heap-overflow vulnerability that gave him control before that authentication function kicked in. Although researchers have uncovered problems with other Samsung devices in the past, Tizen has escaped extensive scrutiny from the security community, probably because it's not widely used on phones yet.

An anonymous reader quotes a report from Ars Technica: Denis Grisak, the man behind the Internet-connected garage opener Garadget, is having a very bad week. Grisak and his Colorado-based company SoftComplex launched Garadget, a device built using Wi-Fi-based cloud connectivity from Particle, on Indiegogo earlier this year, hitting 209 percent of his launch goal in February. But this week, his response to an unhappy customer has gotten Garadget a totally different sort of attention. On April 1, a customer who purchased Garadget on Amazon using the name R. Martin reported problems with the iPhone application that controls Garadget. He left an angry comment on the Garadget community board: "Just installed and attempting to register a door when the app started doing this. Have uninstalled and reinstalled iPhone app, powered phone off/on - wondering what kind of piece of shit I just purchased here..." Shortly afterward, not having gotten a response, Martin left a 1-star review of Garadget on Amazon: "Junk - DO NOT WASTE YOUR MONEY - iPhone app is a piece of junk, crashes constantly, start-up company that obviously has not performed proper quality assurance tests on their products." Grisak then responded by bricking Martin's product remotely, posting on the support forum: "Martin, The abusive language here and in your negative Amazon review, submitted minutes after experiencing a technical difficulty, only demonstrates your poor impulse control. I'm happy to provide the technical support to the customers on my Saturday night but I'm not going to tolerate any tantrums. At this time your only option is return Garadget to Amazon for refund. Your unit ID 2f0036... will be denied server connection."

From an article on FastCompany: Intel hasn't lost its zeal for big leaps in computing, even as it changes the way it introduces new chips, and branches beyond the PC processor into other areas like computer vision and the internet of things. "Number one, too many people have been writing about the end of Moore's law, and we have to correct that misimpression," Mark Bohr, Intel's technology and manufacturing group senior fellow and director of process architecture and integration, says in an interview. "And number two, Intel has developed some pretty compelling technologies ... that not only prove that Moore's law is still alive, but that it's going to continue to provide the best benefits of density, cost performance, and power." But while Moore's law soldiers on, it's no longer associated with the types of performance gains Intel was making 10 to 20 years ago. The practical benefits of Moore's law are not what they used to be. [...] For each new generation of microprocessor, Intel used to adhere to a two-step cycle, called the "tick-tock." The "tick" is where Moore's law takes effect, using a new manufacturing process to shrink the size of each transistor and pack more of them onto a chip. The subsequent "tock" introduces a new microarchitecture, which yields further performance improvements by optimizing how the chip carries out instructions. Intel would typically go through this cycle once every two years. But in recent years, shrinking the size of transistors has become more challenging, and in 2016, Intel made a major change. The latest 14 nm process added a third "optimization" step after the architectural change, with modest performance improvements and new features such as 4K HDR video support. And in January, Intel said it would add a fourth optimization step, stretching the cycle out even further. The move to a 10 nm process won't happen until the second half of 2017, three years after the last "tick," and Intel expects the new four-step process to repeat itself. This "hyper scaling" allows computing power to continue to increase while needing fewer changes in the manufacturing process. If you divide the number of transistors in Intel's current tick by the surface area of two common logic cells, the rate of improvement still equals out to more than double every two years, keeping Moore's law on track. "Yes, they've taken longer, but we've taken bigger steps," Bohr said during his three-hour presentation.

Apple is moving away from the current, cylinder-shaped design used on its Mac Pro desktop, but that replacement will take until next year to hit shelves. From a report: "The Mac Pro, the current vintage that we introduced, we wanted to do something bold and different. In retrospect, it didn't well suit some of the people we wanted to reach," admitted Apple SVP Craig Federighi. "So many of our customers were moving to iMac that we saw a path to address to many, many more of those people," he added. "With the current generation Mac Pro, which some customers love, others may not, one of the things that's certainly clear and true about that is the team tried to do something different, something bold and we always want to encourage the Mac team that whatever products you make, that make customers happy, that we do bold work. Because the Mac's always been about that. It's been about not being conventional thinking, not me-too-stuff," said Phil Schiller. [...] While we'll have to wait until 2018 for the Mac Pro rebirth ("Want to do something great... that will take longer than this year to do," said Schiller), iMac fans can expect a significant update this year, including some new configurations designed specifically for Pro users who already fans of the all-in-one design. [...] Schiller was somewhat less emphatic when I asked if he was willing to make any "courageous" decisions about Mac Pro ports. I thought I saw a little discomfort flicker across Schiller's face as he reacted to that word and he told me that Apple wasn't making promises about ports on the Mac Pro. Port decisions, he said, are made at a product level. "Just because on one product we removed something, doesn't mean we're going to remove it elsewhere," he told me. More on this here.

Apple's expected update to its iMac line will arrive later this year with some previously unexpected additions: pro models. From a report: "We have big plans for the iMac," Phil Schiller, Apple's SVP of worldwide marketing, said during a recent reporter roundtable at the company's Machine Shop hardware prototyping lab. "We're going to begin making configurations of iMac specifically with the pro customer in mind." Just what those configurations will entail, Apple won't yet say. Nor will it comment on the possibility of an iMac Pro moniker for the more powerful machines in the lineup. Company executives are, however, quite happy to confirm a feature the pro iMac will not have: touchscreen. "No," Schiller said when asked if Apple would consider building such a thing. "Touch doesn't even register on the list of things pro users are interested in talking about. They're interested in things like performance and storage and expandability."

A Swedish start-up called Epicenter is offering to implant its employees and start-up members with microchips that function as swipe cards, allowing them to open doors, operate equipment or buy food and drinks with a wave of the hand. While these microchips have been available for decades, the technology has never been implanted in humans on such a broad scale. "Epicenter and a handful of other companies are the first to make chip implants broadly available," reports Associated Press. From the report: [A]s with most new technologies, it raises security and privacy issues. Although the chips are biologically safe, the data they generate can show how often employees come to work or what they buy. Unlike company swipe cards or smartphones, which can generate the same data, people cannot easily separate themselves from the chips. Epicenter, which is home to more than 100 companies and roughly 2,000 workers, began implanting workers in January 2015. Now, about 150 workers have the chips. A company based in Belgium also offers its employees such implants, and there are isolated cases around the world in which tech enthusiasts have tried them out in recent years. The small implants use near-field communication technology, or NFC, the same as in contactless credit cards or mobile payments. When activated by a reader a few inches away, a small amount of data flows between the two devices via electromagnetic waves. The implants are "passive," meaning they contain information that other devices can read, but cannot read information themselves. Ben Libberton, a microbiologist at Stockholm's Karolinska Institute, says hackers could conceivably gain huge swaths of information from embedded microchips. The ethical dilemmas will become bigger the more sophisticated the microchips become. Epicenter workers stage monthly events where attendees can receive the implant.

When the Nintendo Switch was launched in early March, some users complained about dead or stuck pixels, which Nintendo later dismissed as being "normal." We're now approaching the one-month mark and some users are reporting that the console can warp when it's been docked for an extended period of gaming. The Independent reports: [Reddit] User _NSR has posted a picture of a bent Switch online, alongside a message reading, "The Nintendo Switch is starting to warp while only being in dock mode." _NSR expanded on the issue in further comments, explaining, "It does get very hot, considering how small the system is and it is outputting Breath of the Wild for long periods of time on a big screen, it may be too much for it to handle. I'm wondering if it being docked is the problem though, most of my time has been on the dock, and it has to work harder to output to a larger screen. Luckily it hasn't affected the way that it plays, the lack of disc drive is definitely a good thing here." Fellow Reddit users have offered multiple possible explanations for the warp, with battery issues and thermal expansion being suggested. Another Switch user, Magnaha23, backed up _NSR's claim, commenting, "I actually checked my Switch after seeing this. It's starting to do the exact same thing just not as bad as yours yet. I called Nintendo and got a repair set up in like 10 minutes. "I got a prepaid label for overnight shipping. Once they get it it takes a few days to process and repair. If they can repair the unit, which if the whole thing is warped they probably can't, they will send back. Otherwise, they will transfer my data over to a new unit and send that back. They said 6-8 business days once they get my Switch. We shall see..." Other people on the thread, however, say they haven't noticed any warping despite using it for lengthy gaming sessions while docked.

Applehu Akbar writes: Ars Technica reports on a Carnegie-Mellon study of an unexpected side effect of the slowdown in nuclear plant construction after Chernobyl and Three Mile Island. The pollution associated with replacing the power in places where nuclear plants were delayed or canceled has resulted in significantly lower birth weights for children born in the region. The impact on birth weight starts at 97g less in the second quarter after a nuclear shutdown and goes to 146g for in the third quarter, and of similar magnitude thereafter. Though the steady shift in recent years from coal to natural gas has probably slowed this trend down (no update to the study has been announced) because gas pollutes less, Trump's policy of bringing back coal may mean that micro-babies are back in fashion. Here's an excerpt from Ars Technica's report: "[Carnegie Mellon assistant professor of economics and public policy Edson Severnini] looked at the closure of the Browns Ferry nuclear plant in Alabama in 1985 as well as the Sequoyah plant in Tennessee, which was closed from 1985 to 1988. The closure of the two plants corresponded to increased coal burning at nearby coal plants -- in 1985, TVA noted in its annual report that coal plants had 'extraordinary performance' due to the shut down of the nuclear plants. He also gathered birth-weight data from the National Centre for Health Statistics (NCHS) and found that babies born in regions with the biggest increase in coal burning had lower birth weights than babies born in other nearby areas. Looking at data from 1983 to 1985, before the nuclear plant shut down, also showed that the largest change in birth weight occurred after the shutdown."

Doomsday may be closer than ever, but thanks to the Arctic World Archive, at least your data could survive the looming apocalypse. From a report: Norway is already the home to the Global Seed Vault, a frozen ark for 1.5 million seeds to avoid their extinction, and now the Arctic World Archive aims to do the same for your data -- in the same disused mine in the same mountain on the island of Svalbard, famous for its polar bear population. Run by a small Norwegian archiving company called Piql, the World Arctic Archive will store key documents, books and other files on photosensitive film held in protective boxes, a technique Piql says it's tested to survive for at least 500 years and believes will last for 1,000. That longevity is helped by the storage location. More on this here.

From a report on AnandTech: In a bombshell of a press release issued this morning, Imagination has announced that Apple has informed their long-time GPU partner that they will be winding down their use of Imagination's IP. Specifically, Apple expects that they will no longer be using Imagination's IP in 15 to 24 months. Furthermore the GPU design that replaces Imagination's designs will be, according to Imagination, "a separate, independent graphics design." In other words, Apple is developing their own GPU, and when that is ready, they will be dropping Imagination's GPU designs entirely. This alone would be big news, however the story doesn't stop there. As Apple's long-time GPU partner and the provider for the basis of all of Apple's SoCs going back to the very first iPhone, Imagination is also making a case to investors (and the public) that while Apple may be dropping Imagination's GPU designs for a custom design, that Apple can't develop a new GPU in isolation -- that any GPU developed by the company would still infringe on some of Imagination's IP. As a result the company is continuing to sit down with Apple and discuss alternative licensing arrangements, with the intent of defending their IP rights.

An anonymous reader writes: Sunday is World Autism Awareness Day, and landmarks around the world will "light it up blue" as a show of support, including New York's Rockefeller Center and the White House. "Autism spectrum disorders affect an estimated one out of every 68 children in America," President Trump posted Friday, and autistic characters have now even been added to the new Power Rangers movie and on Sesame Street.

But technology could also play a role in improving the live of people with autism spectrum disorders. Reuters is reporting on a robot specifically designed to help teach communication and interaction skills to autistic children, while Vanderbilt University has 20 studies exploring more ways that robotics and technology could help, according to Zachary Warren, an associate professor of pediatrics. "A child may not respond to their mother calling their name but may automatically respond to a robot action or a piece of technology," Warren says after one program which showed improvement in five out of six participants. "If we can use that technology to shift how that child responds, then we may have a very valuable system to that child, that family and maybe for autism intervention."

We all knew this day would come: HTC has introduced a "VR Ad Service" that knows when viewers are actively looking at ads. "Ads that appear in immersive VR environments can not only provide more effective impressions, they can also track whether the users have viewed them or have turned away their gaze. Accordingly, the multiplied effect of effective impressions and verified viewings will bring you higher advertising revenue!" HTC explains. PC Gamer reports: Advertisers will only pay for ads after they've been viewed, according to Business Insider. Some of the formats they will use include loading scenes, 2D and 3D in-app placements, app recommendation banners, and big screen video. This will be an opt-in ad service for developers. HTC notes that by opting in, "all of your free apps would be automatically put on the list which can be used to integrate VR Ads." News of in-game ads coming to VR isn't exactly the sort of thing that will excite gamers. If there's a silver lining here, it's that ads are more likely to be relevant to the viewer's interests over time, at least in theory. "Compared to ordinary ad impressions, ads that are seen by users in a immersive VR environment can not only meet the user's needs by means of precise re-targeting, but can also be detected if they are viewed effectively by users," HTC states. "Therefore, promotion of your applications would have much more effective impression, which not only arouses the attention of potential users and enhance brand image, but further attracts interested users directly to download your apps in the VR environment!"

An anonymous reader writes: "Scientists from two Israeli universities have come up with a way to use flatbed scanners as relay points when sending commands to malware installed on an air-gapped computer," reports BleepingComputer. "Further research also revealed the scanner could also be used to relay stolen data to a nearby attacker. The technique they came up with revolves around the idea that a beam of light could be interpreted as a binary 1 and the lack of visual stimulant can be considered a binary 0."

The attacks can be carried out with lasers mounted on drones, on fixed stands, or by hacking smart light bulbs present near the targeted computer. Attack distances can go up to 900 meters (0.56 miles). During their tests, researchers sent various commands to the PC, such as "d x.pdf" (delete file x.pdf) and "en q" (encrypt folder q). Relaying such commands took between 50 to 100 milliseconds. This research was done by the same team that created methods to steal data from PCs using a hard drive's LED, fan heat, sounds emanated by a computer's GPU fan, electromagnetic signals given out by the GPU, and electromagnetic signals given out by an USB bus.
Here's a PDF of the report, which is titled "Oops!...I think I scanned a malware."

"Lionfish are an invasive species that are destroying our coral reefs and fisheries," writes SkinnyGuy. "The non-profit RISE (from iRobot's Colin Angle) has a plan to use robots to fish these Lionfish and serve them up to us on a delicious, golden platter." Mashable reports: This was not as crazy of an idea as it sounds and Angle had already been wondering "if there was still a way to use robot technology to solve larger environmental problems and maybe more proactively than merely sending our defense robots to natural disaster zones"... Could, Angle wondered, a robot even do the job and could it do it at scale? "Spending half a million dollars to build a robot that kill 10 lionfish is absurd," he told me...

They started with fresh-water electro fishing technology and adapted it for salt water. The robot stuns, but doesn't kill the lionfish and then it sucks them into the robot. It does this over and over again, until full of unconscious fish and then rises to the surface where a fisherman can unload the catch and deliver them to waiting restaurants and food stores. "Ultimately, the control of this device is like a PlayStation game: you're looking at screen and using a joystick controller. Zap it, catch it, do it again, said RISE Executive Director John Rizzi who told me that a team of unpaid volunteers have been working on the prototype for over a year."
The fish-killing robot will launch in Bermuda at the America's Cup festivities on April 19th, where there'll also be a celebrity chef lionfish cook-off and other events to help raise money "to further developer, build and deliver these robots to commercial fishermen and women."

As automation continues to replace human workers, a universal basic income program will begin paying $1,689 per month to select Ontario residents later this year, as Canada joins other countries testing a UBI (which include America, Scotland, the Netherlands, Finland, India, Rwanda, Kenya and Uganda). An anonymous reader quotes the Toronto Star: Public support in Ontario for the province's three-year UBI project to be launched this spring in three Ontario communities is remarkably strong. The 35,000 Ontarians canvassed by Queen's Park for their input were near-unanimous in supporting the UBI projects. And they insisted that a UBI augment, rather than replace, existing welfare, medical and other social supports...

A well-designed UBI equates to freedom. Freedom from exploitative employers. Freedom to launch a small business or develop an invention despite a lack of employment income. Liberation from the "poverty trap," where taking a paying job means surrendering welfare and other benefits... Fact is, job scarcity in traditional vocations is acute, worsening and permanent. In 2013, two Oxford professors forecast that about 45 per cent of U.S. jobs could be eliminated by automation within the next 20 years. And a more recent report by researchers at Indiana's Ball State University found that 88 per cent of U.S. job loss has been caused by automation, not globalization.
Interestingly, the U.S. launched a Universal Basic Income pilot program which ran for three years starting in 1968. It was run by 36-year-old Donald Rumsfeld (who would later become Secretary of Defense) working with special assistant Dick Cheney (who went on to become America's vice president from 2001-2009). U.S. representatives even voted to replace welfare with a UBI, but the measure ultimately failed in the Senate.