Sam’s Club Launching Member Exclusive Event Tomorrow With Discounts on iPad Pro, Apple Watch Series 3, and More
Sam’s Club is launching a member exclusive one day event tomorrow, December 16, with discounts and promotions on the iPad Pro, Apple Watch Series 3 and Series 2, iPhone X, and a few Beats by Dre products. To gain access to the deals, you’ll need a Sam’s Club membership, which start at $45/year.
Note: MacRumors is an affiliate partner with Sam’s Club. When you click a link and make a purchase, we may receive a small payment, which helps us keep the site running.
The sale items will be available to purchase in your local Sam’s Club store (beginning at 7:00 a.m. ET) as well as online (beginning at 12:01 a.m. ET), but the retailer noted that there are limited quantities on all items. Some products will see discounts only lasting December 16, while others will run through Christmas.
Below are a few of the Apple products and accessories with promotions or deals at Sam’s Club tomorrow (note that the Beats Solo 3 deal is already up):
- iPhone X – Get a $150 Sam’s Club Gift Card with activation on installment: AT&T, Verizon and USCell
- 12.9-inch and 10.5-inch iPad Pro, 256GB – $125 off starting 12/16
- Apple Watch Series 3 and Series 2 – $25 off only on 12/16
- Powerbeats3 Wireless – $128 only on 12/16 (matching Amazon and Walmart)
- Beats Solo3 – $198 through 12/25 (compared to $240 at Amazon and $220 at Walmart)
- $50 iTunes Card – $42.98
- $100 iTunes Card multipack – $84.47
Although $25 off isn’t the best deal we’ve seen for Apple Watch Series 2 models this season, it is one of the first straight cash discounts on Series 3, which as of yet have only received promotions for store credit of some kind, like at Kohl’s on Black Friday. This should knock down the non-cellular Aluminum Series 3 devices to about $305 for 38mm and $335 for 42mm. Cellular models in Aluminum will be $375 for 38mm and $405 for 42mm.
Check out the catalog for the December One Day Only event at Sam’s Club for a longer list of products, and then head over to our Deals Roundup for more holiday sales.
Related Roundup: Apple DealsTag: Sam’s Club
Discuss this article in our forums
Employees at a Japanese firm will soon be able to receive part of their salary in Bitcoin, if they’re feeling brave enough. GMO Internet, which offers a range of web-related services including a Bitcoin exchange, will pay workers up to 100,000 yen (about $890) starting in February. “Employees can receive salaries by Bitcoin if they want,” a company spokesperson said. “We hope to improve our own literacy of virtual currency by actually using it.”
The venture is strictly voluntary, and the lower limit will be 10,000 yen, around $88. The company is also offering a 10 percent bonus for employees willing to try it, which might help make up for some of the crypto-currency’s extreme volatility.
GMO plans its own initial coin offering (ICO), with a strange twist. The company’s “GMO coin” will be a token that buyers can use toward buying not only Bitcoins, but next-gen mining boards or part of a cloud mining business. The company plans to develop “cutting-edge 7-nanometer process technology” that will allow high performance mining with lower power consumption than current chips.
Bitcoin is functioning less as a currency and more as an investment nowadays, as we wrote recently, having dramatically soared in value from around $800 to as high as $19,000 in a year. Employees willing to take them as pay would probably therefore sock them away rather than using the currency to buy snacks. It’ll be an interesting experiment, in any event, especially in a country that has taken a forward-looking approach to the currency, and hosted one of its biggest failures.
Via: The Guardian
Source: GMO (translated)
Simon Stålenhag’s illustrated novel The Electric State is coming to the big screen with It director Andy Muschietti likely at the helm. Russo Brothers Studio beat at least three other studios for rights to the project, Deadline reports, and a number of producers and directors sought to be involved.
The narrative artbook was funded through Kickstarter and depicts a bleak US. The book follows a runaway teenager and her toy robot as they travel through the country. Remnants of large battle drones and piles of trash from a declining high tech society are strewn about the countryside and as the teen and her robot travel towards the west coast, things keep getting worse and worse.
Joe and Anthony Russo directed the last two Captain America movies, the writers of which — Christopher Markus and Stephen McFeely — will be writing The Electric State’s script. The Russo brothers are also directing the two The Avengers sequels. In a statement to Deadline the Russos said, “The opportunity to partner with inspirational talents like Simon, Andy, Chris and Steve is exactly the reason we started our company. We can’t wait to help this team create something special.”
By Signe Brewster
This post was done in partnership with Wirecutter, reviews for the real world. When readers choose to buy Wirecutter’s independently chosen editorial picks, it may earn affiliate commissions that support its work. Read the full article here.
After spending 35 hours researching and testing seven of the best kits for learning robotics, we found the Lego Boost to be the best kit for most beginners. With its Lego-based design, built-in sensors, and the most expansive set of options for creativity and personalization, it was the most fun to build with. And the streamlined tablet app’s user-friendly instructions and super-simple programming made it the easiest to learn of any of the kits we tried.
Who should get this
A robotics kit should be of interest to any parent who wishes to encourage a child’s interest in the STEM fields. Schools often turn to kits to teach physics or math concepts, but at home they can also serve as a creative outlet and source of fun. We think the best beginner robotics kits challenge you to build the robot in a creative, open-ended way using hundreds of pieces. You’ll then program it on your mobile device or computer to walk, talk, and play games.
How we picked and tested
We interviewed experts and scoured sites like Amazon for robotics kits before selecting seven to test. Photo: Signe Brewster
To choose which beginner robotics kits to test, we scoured websites like Amazon for reviews and considered existing guides from publications such as Tom’s Guide and Make. We also consulted three experts on the most important functions to look for in an introductory robotics kit. We eventually settled on seven kits to test, based on these criteria:
- Encourages both building and programming the robot
- Includes everything you need in the box
- Thorough, easy-to-follow instructions
- Open-ended design, and easily expandable with optional kits
- Intuitive graphical programming software (not text-based)
- Priced between $85 and $300
For each kit, we began by following the included instructions to build one of the suggested models. We timed the process, noting any difficulties we encountered and how much fun we had. Then, we downloaded the required app or desktop software, and went through the steps to build a basic program. We considered the difficulty level of each coding language, and if it would satisfy builders both young and old.
Our pick: Lego Boost
Photo: Signe Brewster
The Lego Boost robotics kit offers the best overall experience for those with no programming experience. The 847-piece kit is made from Lego pieces, making it the most fun and the easiest to build among kits of its size, thanks to Lego’s familiarity and versatility. And the fact that any Lego block can be used in a project creates a massive opportunity for expansion right out of the box.
The instructions for building and programming the robot are provided via a simple tablet app that even nonreaders can follow. Its programmable sensors can detect movement, distance, and color. However, the Lego Boost app is tablet-only, so it won’t work with a smartphone.
Making an easy-to-use robot comes at a price. Lego Boost is one of the most expensive kits we tested, but it’s also the most expansive for building right out of the box—especially if you have other Lego kits at home.
Runner-up: Jimu AstroBot
Photo: Signe Brewster
If the Lego Boost kit is unavailable, we recommend the Jimu AstroBot kit. It’s a smaller kit that comes with three suggested builds, two fewer than the Boost, and its sensor is less sophisticated—it can detect objects, but not color or movement like Boost’s. It also costs more, and because it isn’t Lego, it doesn’t offer the same amount of expansion possibilities. AstroBot is just as simple to put together and program as the Boost, but it also includes hands that can grip objects and eyes that can be programmed to show different colors and patterns. It’s also one of the cutest robots we built.
AstroBot’s app is the best we tried. There’s a “story” mode that adds a plot to the curriculum meant to teach the basics of the robot, which makes building the robot more appealing for kids. You can also jump straight into more traditional directions. We thought controlling AstroBot’s hands and eyes made the programming stage particularly exciting.
Budget pick: Robotis Ollobot Play 700
Photo: Signe Brewster
The Robotis Ollobot Play 700 is one of the least expensive kits we tested, but it also offers one of the widest ranges of programming options—from Robotis’s version of Scratch to a more advanced C++ environment—which means the robot’s abilities can grow with your level of experience. The software is compatible with iOS and Android devices (both phones and tablets) and Windows PCs, though we ran into some issues getting the Android app to connect properly.
The Robotis was one of the simplest kits we attempted, with only 233 pieces and four suggested builds. However, each suggested build has some unique features, and the robots can follow courses and respond to stimulus such as clapping.
Upgrade pick: Vex IQ Super Kit
Photo: Signe Brewster
The Vex IQ Super Kit is one of the most expensive kits we tested. For roughly twice the price of the Lego Boost kit, you get more advanced building and programming features that will be appealing if you want to get deeper into robotics, as well as ample documentation. Vex also offers a suite of programming options, including an iPad app and desktop Scratch and C-based desktop options.
Vex provides a lot of context for its eight suggested builds, including explanations of different build steps and the purpose of each sensor. You come out of the process knowledgeable about the name and explicit purpose of each part of the robot, something we can’t say is true with the Lego Boost.
This guide may have been updated by Wirecutter. To see the current recommendation, please go here.
Note from Wirecutter: When readers choose to buy our independently chosen editorial picks, we may earn affiliate commissions that support our work.
Today SpaceX launched its first reused rocket for NASA. Both the Falcon 9 rocket itself and the Dragon capsule have been used prior to this resupply mission. Back in June, Elon Musk’s spacefaring venture put a reused capsule in orbit, but this trip to the International Space Station marks the first time that NASA has used a “flight proven” booster on a mission (read: reused) according to CNBC. Based on the livestream, everything seems to have gone well: SpaceX successfully landed the rocket, which means it could potentially make its third flight in the future.
The Dragon will spend around a month attached to the ISS and will return to Earth on January 13th. If you missed any of today’s action, NASA is replaying the launch stream ahead of today’s press conference at noon Eastern.
Rocket and spacecraft for CRS-13 are flight-proven. Falcon 9’s first stage previously launched SpaceX’s eleventh resupply mission for @NASA, and Dragon flew to the @Space_Station in support of our sixth cargo resupply mission. pic.twitter.com/RY4F2TrWO2
— SpaceX (@SpaceX) December 6, 2017
In late October, Chair of the Digital, Culture, Media and Sport Select Committee Damian Collins called upon Facebook and Twitter to report back with any evidence of Russian meddling in the UK’s vote to leave the EU. It’s an evolution of the committee’s inquiry into the problem of fake news, which Collins considers a pretty serious “threat to democracy.” Responses from Facebook and Twitter have landed this week, and if you haven’t been following along, it’s safe to say Collins isn’t particularly impressed with how deep the social networks are digging to identify the true scale of political misinformation and influence exerted by Russia.
Confusion around what’s expected of Facebook and Twitter seems to stem from the fact there are two different, ongoing investigations into EU referendum interference happening in the UK right now. Collins’ committee is looking at the big picture: How fake news, Twitter bots, promoted posts and everything in between may’ve swayed public opinion one way or another. The Electoral Commission’s independent inquiry is casting a tighter net, primarily concerned with ad spending on social platforms. This is part of a wider initiative by the commission to create greater transparency for online political ads, so voters know what articles, posts or digital campaigns with a political angle are actually funded content, and who’s stumping up the cash for them.
Earlier this week, Facebook concluded that the Internet Research Agency (IRA) put hardly any effort into influencing public opinion ahead of the Brexit vote. The IRA, sometimes referred to as the “Russian troll farm,” is an organisation headquartered in St. Petersburg that’s said to operate on behalf of the Russian government. It is accused of using fake accounts and bots on social media, news sites, blogs, forums and the like to influence debates on all kinds of political matters, with the goal of championing the Kremlin’s interests. Facebook says that pages and profiles linked to the IRA spent a total of $0.97 on three ads, which ran for four days in May 2016 (prior to the June Brexit vote). These ads “were also targeted to US audiences and concerned immigration, not the EU referendum,” and resulted in “approximately 200 impressions to UK viewers.”
Twitter has also responded, relaying info on “six referendum-related ads” that ran from April to June last year. These were purchased for $1,031.99 by RT (Russia Today), a state-sponsored news agency. Whether these promoted posts should be considered as ‘interference’ is up for discussion. RT is a recognised media outlet with a pretty well-known, Russia-leaning bias, and the tweets themselves point to articles on the referendum vote, but aren’t headlined with any obvious agenda. It’s also not uncommon for a news outlet to promote content on a hot topic. In other words, not exactly the type of clandestine campaign the IRA is infamous for.
In October this year, Twitter made the call to ban RT and Sputnik (another Russia-controlled news agency) from buying ads on the platform. Both outlets are accused of being propaganda machines that interfered with the 2016 US presidential election. Just a few weeks ago, Google took similar steps, announcing it would start downranking RT and Sputnik in its search results. For what it’s worth, Google has also responded to the inquiries in the UK with the one-liner: “We took a thorough look at our systems and found no evidence of this activity on our platform.”
Acknowledging the responses it’s received, the Electoral Commission said: “Facebook, Google and Twitter have responded to us. We welcome their cooperation. There is further work to be done with these companies in response to our request for details of campaign activity on their platforms funded from outside the UK. Following those discussions we will say more about our conclusions.”
That’s about as good as it gets from the Electoral Commission, which doesn’t typically comment on ongoing inquiries. Collins is a bit more forthcoming in his reactions, and he’s clearly unimpressed with the one-size-fits-all responses given to both the commission and his committee, which are looking at potential Russian influence through slightly different lenses.
Summarising his tweetstorm on Facebook’s response, he says that the platform only looked into accounts and pages previously identified as IRA-run during a similar investigation into US election meddling. “Facebook conducted research to identify 10,000s of fake pages and accounts that were active during the French Presidential election. They should do the same for the EU referendum, and not just rely on external sources referring evidence of suspicious activity back to them.” He’s met with Facebook and asked it conduct a more thorough examination of both “Russian activity linked to the 2017 General Election, as well as the EU referendum.”
Twitter hasn’t fared much better. In a letter to CEO Jack Dorsey, Collins called the social network’s response “completely inadequate.” He’s demanded more information (and not for the first time) on bots and accounts linked to the IRA that were potentially used for “foreign interference” in UK political matters. He goes on to say his committee are looking well beyond paid ads (the domain of the Electoral Commission’s investigation): “It seems odd that so far we have received more information about activities that have taken place on your platform from journalists and academics than from you.”
Collins refers specifically to a study by London’s City University that suggests a “13,500-strong Twitter bot army” that posted nearly 65,000 tweets with a “clear slant towards the leave campaign” disbanded after the vote. These accounts were either deleted “or were suddenly blocked or removed by Twitter” following the referendum, and more than 25,000 other accounts changed usernames abruptly. Another study by the University of Edinburgh spotted more than 400 IRA-operated accounts that got involved in the Brexit debate by looking through the usernames Twitter supplied to US authorities as part of their election interference investigations.
Collins is frustrated, in part, by the level of cooperation Facebook, Twitter and others have shown to US officials, not to mention the proactive steps taken by Facebook specifically to suppress the impact fake news might have on elections in Germany and France. The saga is far from over, of course, and we expect to see further dialogue between Facebook, Twitter and the select committee over the coming weeks and maybe months — assuming the social media giants find conclusive evidence of additional Russian activity, that is.
Despite even Prime Minister Theresa May throwing around accusations, there’s a chance both ongoing inquiries could conclude anticlimactically. The Oxford Internet Institute is due to publish a study any day now that claims Russian interference in the Brexit vote was fairly negligible. Speaking to The New York Times, one of the study’s authors Philip Howard said: “Overall, I think Russian activity during Brexit seems to have been minimal. The real source of misinformation about the Brexit debate was homegrown.”
There’s also no guarantee we’ll ever know the full extent to which noise on the internet shaped public opinion, since it’s a pretty complex issue to quantify. And while Russia is taking the brunt of accusations right now, it’s by no means the only culprit. A recent report from Freedom House claims that governments in around 30 countries are “mass producing their own content to distort the digital landscape in their favor,” with elections in 18 countries being in some way affected by this form of online influencing.
Images: Getty Images (RT microphone); Sputnik Photo Agency / Reuters (Vladimir Putin); PA Wire / PA Images (Brexit bus)
Resellers have existed since way before the internet. But with the surge of online shopping, they have found the perfect weapon to aid their business: bots, automated software that can add products to a virtual cart and purchase them faster than any human. Even if your browser autofills personal information, like your address and credit card, and it only took you one minute to get to the checkout page — that’s too slow. Bots can buy almost anything in a matter of seconds, which is why they’ve become the ideal tool for people who make money by snatching up coveted items and selling them on eBay, Amazon and Craigslist for a profit.
With the holiday-gifting season just around the corner, the effect these bots have on the market is crystal clear. One of the hottest children’s toys of the season, Fingerlings (a strangely cute monkey that you can wrap around your finger), is sold out at all major retailers, including Amazon, Best Buy, Target, Toys R Us and Walmart. And every time they restock, resellers snatch them up within seconds. Fingerlings, which come in a variety of colors, retail for around $15, but the cheapest you can find them for at the moment on sites like eBay is about $40 — more than double their original price.
You’ll find a similar case with Nintendo’s SNES Classic Edition. Although it’s been months since it launched, the mini retro console is also out of stock everywhere online. And, like with the Fingerlings, every time it restocks on GameStop or another retailer, it’s gone almost instantly. The SNES Classic Edition has an MSRP of $80, but on the reseller market it goes for anywhere between $115 to $200. This is partially why Amazon, Best Buy and others have resorted to selling it at brick-and-mortar stores instead, where drops are random and more easily limited to one per customer.
Stores won’t share data to reveal how many of these go to bots, but a quick look at people’s tweets of disappointment suggests that resellers grabbed whatever stock was loaded. What happens is that a retailer will tweet that the SNES Classic is available again in limited quantities, but by the time most people add the product to the cart and go to check out it’s no longer available. Resellers can successfully beat this system by creating a software script that constantly scans product pages and sees when they come back in stock even before the store can tell the general public.
While bots are at the core of a reseller’s system, there’s also a network of Twitter accounts and stock-monitoring sites that help them get to any product first. Steal Supply, for example, only has about 7,000 followers on Twitter, but the account is one of the most effective when it comes to tweeting about restocks of popular products, such as the SNES Classic. There are many accounts and services similar to Steal Supply, created to help not just resellers but regular shoppers buy sought-after stuff. Nowinstock.net, for example, lets you make an account and get product stock alerts sent directly to your email or phone.
Sometimes all you need is to follow a Twitter account, or a website, and have push notifications turned on. Catching them off guard, and being on high alert, might be the best way to potentially beat the bots. It may not be the ideal shopping experience, but remember all the things Arnold Schwarzenegger’s character went through in Jingle All the Way to get a Turbo-Man toy for his kid? Things could be much worse.
Nike has taken an interesting approach to fight software scripts designed for shopping. The sportswear giant recently started using augmented reality to sell limited-edition, highly coveted sneakers. Through its SNKRS app, users can go to a specific place, look for a physical sticker and use their phone to bring up a 3D model of the shoe. If you find it, you can buy right there and then. This means an actual human is necessary for the entire purchase process, but not everyone is going to have a compatible device. It’s also not exactly different from brick-and-mortar sales, since you have to physically go to a place to buy the shoe.
Nike’s secret weapon against bots: AR.
The real solution could be legislation, the kind that would outlaw the use of bots in online shopping altogether. Last year, former President Barack Obama signed the “Better Online Ticket Sales (BOTS) Act of 2016” into law, which made it illegal for individuals to use bots (or other similar computer software) to buy more tickets than a specified limit to concerts, sports games and other public events.
According to the BOTS Act, it is a violation of the law to “circumvent a security measure, access control system, or other technological control or measure on an internet website or online service that is used by the ticket issuer to enforce posted event ticket purchasing limits or to maintain the integrity of posted online ticket purchasing order rules.” What’s more, it’s now also an offense to knowingly buy a ticket to an event from someone who used a bot.
Democratic New York Senator Chuck Schumer, who helped introduce the original BOTS Act bill in 2016, did not answer our request for comment. But he put out a statement earlier this month urging retailers to find ways to block bots. “Grinch bots cannot be allowed to steal Christmas, or dollars, from the wallets of New Yorkers,” he said. “Middle class folks save up — a little here, a little there — working to afford the hottest gifts of the season for their kids but ever-changing technology and its challenges are making that very difficult. It’s time we help restore an even playing field by blocking the bots”
Senator Chuck Schumer at a press conference.
Schumer added that when it comes to online shopping, “major retailers should put forth policies that will help prevent future Grinch bots from stealing the season’s hottest toys.” In 2016, he said that the anti-scalper legislation was inspired by the inflated ticket prices for the popular musical Hamilton, and he believes the “same cyber scalpers” have now moved on to toys and other products that are high in-demand during the holiday season. At the early stages of the BOTS Act, it was reported that resellers had a feast and bought over 20,000 Hamilton tickets to sell them for as much as four times their face value.
As it stands, Amazon, Best Buy and others don’t seem to have a good answer for bots, other than limiting the amount of items people can buy with one account — and sometimes one credit card or shipping address. The problem with the only-one-item-per-account system is that resellers can easily get around that by having multiple accounts, each filled with different personal information that won’t trigger any alarms. The other option is to simply sell products at a physical store, where a reseller’s “bots” are just friends and family that wait in line with them.
A source inside Amazon confirmed to Engadget that the company’s main way to stop bots is to limit purchases of high-demand products. Additionally, the source said that if third-party sellers on its site don’t price items competitively, the company can always remove their ability to sell them. Still, for now, it seems like the only way to really stop them is for the US Government to introduce a similar bill to the BOTS Act. Though the law could get complex if it needs to cover anything and everything sold online.
Grinch bots cannot be allowed to steal Christmas, or dollars, from the wallets of New Yorkers. It’s time we help restore an even playing field by blocking the bots.
Senator Chuck Schumer
If this continues to be a problem for retailers and customers start to lose faith in them, they could lobby US representatives and senators to make the case for broader legislation.
Until that happens though, your best bet to get that SNES Classic Edition or Fingerlings in time for the holidays may be to build a bot yourself — and do you really want to do that and potentially make matters worse? For those who aren’t savvy enough but want to know what being a Grinch bot is about, there are services out there that sell you a computer script to help with your chase, however even they don’t guarantee that you’ll get the item you want.
Either of these options sound like a nightmare, because it shouldn’t be this complicated to buy a Christmas present. But that’s what you can do right now, unless of course you don’t mind paying those ridiculously inflated prices.
Facebook launched its Facebook 360 app for Gear VR in March and today the company announces some new features. Facebook says that because people want more than just 360 videos in VR, it’s adding a slew of new video types to the app’s available options.
Now, users will be able to view Live videos, including Live 360 videos rendered in 4K, shows available through Facebook Watch, gaming and eSports videos, videos you’ve previously watched on Facebook as well as videos and 360 photos you’ve uploaded or shared through Facebook. The app will also now allow you to search for and follow Pages.
Facebook has been expanding its VR offerings this year. In April, it launched Spaces, its social VR experience and in October, Mark Zuckerberg announced the company was working on a VR feature that would let you meet up with friends and watch concerts. Creating a wider viewing selection in Facebook 360 is a useful addition and can only help boost the company’s VR ambitions.
This is a bit ironic. When FCC chairman and former Verizon lawyer Ajit Pai uploaded a video this week detailing all the things we’ll still be able to do after he killed net neutrality, he apparently forgot one key step: asking permission. You see, part of the smarmy and glib clip the FCC produced with The Daily Caller has EDM producer Baauer’s “Harlem Shake” in it. Baauer claims he received no licensing requests for the once-viral track and in a tweet pledged that “I’m taking action. Whatever I can do to stop this loser.”
In a statement to Billboard, Baaur said:
“The use of my song in this video obviously comes as a surprise to me as it was just brought to my attention. I want to be clear that it was used completely without my consent or council. My team and I are currently exploring every single avenue available to get it taken down. I support Net Neutrality like the vast majority of this country and am appalled to be associated with its repeal in anyway.”
However, this could be seen as “fair use.” Fair use means a portion of of a work can be used without obtaining permission from the copyright holder. The section of the FCC’s video with “The Harlem Shake” is under 20 seconds long. More than that, the video could be protected by parody laws given how the admittedly un-funny clip is structured. It’s all up to how a judge will interpret the case in front of them.
“A parody is a work that ridicules another, usually well-known work, by imitating it in a comic way,” Stanford’s Rich Stim writes. “Judges understand that, by its nature, parody demands some taking from the original work being parodied. Unlike other forms of fair use, a fairly extensive use of the original work is permitted in a parody in order to ‘conjure up’ the original.” It’s how “Weird” Al Yankovic has a career, regardless of the fact that he asks permission before writing his takes on pop songs.
So, while it might be infuriating, there’s a chance that Pai didn’t do anything wrong here. His actions yesterday are another matter entirely, of course.
Tech’s biggest players have fully embraced the AI revolution. Apple, Qualcomm and Huawei have made mobile chipsets that are designed to better tackle machine learning tasks, each with a slightly different approach. Huawei launched its Kirin 970 at IFA this year, calling it the first chipset with a dedicated neural processing unit (NPU). Then, Apple unveiled the A11 Bionic chip, which powers the iPhone 8, 8 Plus and X. The A11 Bionic features a neural engine that the company says is “purpose-built for machine learning,” among other things.
Last week, Qualcomm announced the Snapdragon 845, which sends AI tasks to the most suitable cores. There’s not a lot of difference between the three company’s approaches — it ultimately boils down to the level of access each company offers to developers, and how much power each setup consumes.
Before we get into that though, let’s figure out if an AI chip is really all that much different from existing CPUs. A term you’ll hear a lot in the industry with reference to AI lately is “heterogeneous computing.” It refers to systems that use multiple types of processors, each with specialized functions, to gain performance or save energy. The idea isn’t new — plenty of existing chipsets use it — the three new offerings in question just employ the concept to varying degrees.
The Snapdragon 845.
Smartphone CPUs from the last three years or so have used ARM’s big.LITTLE architecture, which pairs relatively slower, energy-saving cores with faster, power-draining ones. The main goal is to use as little power as possible, to get better battery life. Some of the first phones using such architecture include the Samsung Galaxy S4 with the company’s own Exynos 5 chip, as well as Huawei’s Mate 8 and Honor 6.
This year’s “AI chips” take this concept a step further by either adding a new dedicated component to execute machine-learning tasks, or, in the case of the Snapdragon 845, using other low-power cores to do so. For instance, the Snapdragon 845 can tap its digital signal processor (DSP) to tackle long-running tasks that require a lot of repetitive math, like listening out for a hotword. Activities like image recognition, on the other hand, are better managed by the GPU, Qualcomm’s director of product management Gary Brotman told Engadget. Brotman heads up AI and machine learning for the Snapdragon platform.
Meanwhile, Apple’s A11 Bionic uses a neural engine in its GPU to speed up Face ID, Animoji and some third-party apps. That means when you fire up those processes on your iPhone X, the A11 turns on the neural engine to carry out the calculations needed to either verify who you are or map your facial expressions onto talking poop.
On the Kirin 970, the NPU takes over tasks like scanning and translating words in pictures taken with Microsoft’s Translator, which is the only third-party app so far to have been optimized for this chipset. Huawei said its “HiAI” heterogeneous computing structure maximizes the performance of most of the components on its chipset, so it may be assigning AI tasks to more than just the NPU.
Differences aside, this new architecture means that machine learning computations, which used to be processed in the cloud, can now be carried out more efficiently on a device. By using parts other than the CPU to run AI tasks, your phone can do more things simultaneously, so you are less likely to encounter lag when waiting for a translation or finding a picture of your dog.
Plus, running these processes on your phone instead of sending them to the cloud is also better for your privacy, since you reduce the potential opportunities for hackers to get at your data.
The A11 Bionic’s two “performance” cores and four “efficiency” cores.
Another big advantage of these AI chips is energy savings. Power is a precious resource that needs to be allocated judiciously, since some of these actions can be repeated all day. The GPU tends to suck more juice, so if it’s something the more energy efficient DSP can perform with similar results, then it’s better to tap the latter.
To be clear, it’s not the chipsets themselves that decide which cores to use when executing certain tasks. “Today, it’s up to developers or OEMs where they want to run it,” Brotman said. Programmers can use supported libraries like Google’s TensorFlow (or more specifically its Lite mobile version) to dictate on which cores to run their models. Qualcomm, Huawei and Apple all work with the most popular options like TensorFlow Lite and Facebook’s Caffe2. Qualcomm also supports the newer Open Neural Networks Exchange (ONNX), while Apple adds compatibility for even more machine learning models via its Core ML framework.
So far, none of these chips have delivered very noticeable real-world benefits. Chip makers will tout their own test results and benchmarks that are ultimately meaningless until AI processes become a more significant part of our daily lives. We’re in the early stages of on-device machine learning being implemented, and developers who have made use of the new hardware are few and far between.
Right now, though, it’s clear that the race is on to make carrying out machine learning-related tasks on your device much faster and more power-efficient. We’ll just have to wait awhile longer to see the real benefits of this pivot to AI.
Images: Huawei (Kirin AI processor), Apple (A11 processor cores).