-->
UnlabelledTechnology - Google News

Thursday, September 3, 2020

author photo

Technology - Google News

Super Mario Bros. 35th Anniversary Direct - Nintendo

Posted: 03 Sep 2020 06:00 AM PDT

How Google Maps uses DeepMind’s AI tools to predict your arrival time - The Verge

Posted: 03 Sep 2020 07:00 AM PDT

Google Maps is one of the company's most widely-used products, and its ability to predict upcoming traffic jams makes it indispensable for many drivers. Each day, says Google, more than 1 billion kilometers of road are driven with the app's help. But, as the search giant explains in a blog post today, its features have got more accurate thanks to machine learning tools from DeepMind, the London-based AI lab owned by Google's parent company Alphabet.

In the blog post, Google and DeepMind researchers explain how they take data from various sources and feed it into machine learning models to predict traffic flows. This data includes live traffic information collected anonymously from Android devices, historical traffic data, information like speed limits and construction sites from local governments, and also factors like the quality, size, and direction of any given road. So, in Google's estimates, paved roads beat unpaved ones, while the algorithm will decide it's sometimes faster to take a longer stretch of motorway than navigate multiple winding streets.

All this information is fed into neural networks designed by DeepMind that pick out patterns in the data and use them to predict future traffic. Google says its new models have improved the accuracy of Google Maps' real-time ETAs by up to 50 percent in some cities. It also notes that it's had to change the data it uses to make these predictions following the outbreak of COVID-19 and the subsequent change in road usage.

"We saw up to a 50 percent decrease in worldwide traffic when lockdowns started in early 2020," writes Google Maps product manager Johann Lau. "To account for this sudden change, we've recently updated our models to become more agile — automatically prioritizing historical traffic patterns from the last two to four weeks, and deprioritizing patterns from any time before that."

The models work by dividing maps into what Google calls "supersegments" — clusters of adjacent streets that share traffic volume. Each of these is paired with an individual neural network that makes traffic predictions for that sector. It isn't clear how large these supersegments are, but Googles notes they have "dynamic sizes," suggesting they change as the traffic does, and that each one draws on "terabytes" of data. The key to this process is the use of a special type of neural network known as Graph Neural Network, which Google says is particularly well-suited to processing this sort of mapping data.

For more detail, check our the blog posts from Google and DeepMind here and here.

Let's block ads! (Why?)

Impressions from Intel’s Tiger Lake laptop CPU launch event - Ars Technica

Posted: 03 Sep 2020 04:47 AM PDT

Intel shows off its platform work on motherboard miniaturization—which is key to modern, sleek, ultrathin designs that won't allow stacking components on <em>top</em> of the board anymore.
Enlarge / Intel shows off its platform work on motherboard miniaturization—which is key to modern, sleek, ultrathin designs that won't allow stacking components on top of the board anymore.
Intel
Intel held a launch event today for its next-generation laptop CPU family, codenamed Tiger Lake. There wasn't much new information about Tiger Lake itself, though—if you followed our coverage of Intel Architecture Day last month, you already know most of the technical detail covered at today's event.

Intel's story on the raw performance of Tiger Lake today holds constant with both what the company announced at Architecture Day, and what the leaked i7-1185G7 benchmarks implied—significantly higher performance from the i7-1185G7 than from AMD's Ryzen 7 4800U, in both CPU and GPU performance.

Taking direct aim at Renoir

  • Intel took very specific aim at a clearly named target today: the Ryzen 7 4800U.
    Intel
  • Intel took great pains to declare large-margin victories over the Ryzen 7 4800U on tests run on a wide array of real-world applications.
    Intel
  • The most compelling presentations showed split screen video with a Ryzen 4800U powered laptop on the left, and an Intel i7-1185G7 on the right. 4K video playback in Premiere stuttered badly on the Ryzen side.
    Intel
  • A streaming presentation in which a creator gave an Adobe Lightroom tutorial on Ryzen 4800U (left) vs i7-1185G7 (right) showed a lot of spinning wait circles on the Ryzen side, and none on the Intel side.
    Intel

We did see considerably more direct discussion of that competitive performance, however, with some pretty compelling side-by-side video of gaming, Adobe Premiere, and other tasks to back up Intel's claims of market performance leadership with the upcoming parts. Of course, Intel has more angles to play here than raw hardware performance—the company has software partnerships with vendors like Adobe to make certain that its proprietary "value-added" features like Deep Learning Boost (aka AVX-512) are leveraged by those vendors.

In particular, the Adobe Premier, Photoshop, and Lightroom demonstrations leaned on AI-powered features using the Intel OpenVINO platform to perform inference workloads, taking advantage of Intel AVX-512 instructions. On the one hand, this is "unfair" to AMD—on the other hand, we're not so certain that matters much to someone whose workload is largely Adobe Premier or other applications where Intel has gotten a software partnership foothold with the vendor.

Faster is faster, and slower is slower—as Intel manages to get the utilization of features like AVX-512 out into the wider application market, AMD will need to figure out a strategy to adapt and compete.

Deep marketing focus on Project Athena

  • Intel's new "Evo" branding requires an Intel i5 or i7 CPU—but it also requires a lot of directly user-facing "experience" features.
    Intel
  • Half the launch event was about the Athena/Evo platform branding, not Tiger Lake itself. Much of that revolved around the market research Intel did to try to understand how users use their machines, and what they really want.
    Intel
  • Project Athena itself is about user experience metrics, such as battery life, slim design, and fast wake. These platform planks are the underlying hardware challenges that must be overcome to enable those user "experiences."
    Intel
  • Intel's market research focused on four user verticals when deciding what needs and use cases Project Athena needed to guarantee answers for.
    Intel

Apart from the side-by-side video demonstrating Tiger Lake's high performance, the most interesting part of the launch event wasn't really about Tiger Lake at all—it was about Intel's Project Athena laptop certification and verification platform, and its newest branding "Evo." A full half of the all-day presentation was devoted entirely to Athena and Evo, with very little mention of the actual hardware underneath. Instead, Intel wanted to convey a message of researching, listening to, and adapting to how end customers use their laptops.

Athena and its subset Evo aim to create a guaranteed, branded level of user experience—with minimum values for targets, such as long battery life, light-weight screen brightness, rapid wake time, and so forth. Although you can't call a non-Intel laptop "Evo"—the specification requires a Core i5 or Core i7 CPU—Intel's marketing works hard to frame Tiger Lake as more of a way to achieve the whole-system user experience that Athena and Evo guarantee than as a fully fledged product in its own right.

Conclusion

We're hesitant to make any firm proclamations about hardware we've only seen a few limited videos of—Intel has had a very rough couple of years, its marketing hasn't always been the most accurate, and it desperately needs a win here. Its side-by-side "against the competition"—meaning Ryzen 4800U—videos are quite compelling, but we'll need direct third-party testing to see just how much of the advantage shown might require an artificially narrow workload.

What we are pretty confident about is that Tiger Lake looks like a much more meaningful competitor for AMD than Intel has been able to field for the last few cycles. Intel credits its underlying SuperFIN technology for the majority of the improvements, and that technology will apply to new desktop CPU designs as well—so if Tiger Lake pans out well, we can expect to see a similar renaissance in Intel's desktop CPUs in 2021.

We hope to get our hands on one or more Tiger Lake-powered laptops for independent testing and review in November, if not sooner.

Let's block ads! (Why?)
Next article Next Post
Previous article Previous Post

This post have 0 komentar


EmoticonEmoticon

Next article Next Post
Previous article Previous Post