-->

Thursday, August 16, 2018

author photo

Technology - Google News


Android 9 Pie review: the predictive OS

Posted: 16 Aug 2018 06:00 AM PDT

Here's the story with every new version of Android, in a nutshell: it's great, but you can really only get it on a phone that Google makes. Sometime next year, new phones by other companies will launch with it. The Android phone in your pocket might get it, maybe, but it'll take longer than you want, and honestly, the new version isn't that different, so you shouldn't sweat it too much. Yes, fragmentation is an issue, but it's better now than it used to be, thanks to Google's ability to push some key updates out through the Google Play store instead of having to rely on full system updates.

The story with Android 9 Pie isn't radically different, but it changes some of those tried and true (and increasingly tired) lines a bit. For the first time, I've had a chance to test the official release of a new version of Android on a phone not made by Google, the Essential Phone. That's a good sign.

Although a few of the promised features aren't shipping or are still in beta, I think this version of Android is good enough that users should demand the update for their phones. I'm not trying to organize a campaign to shake off our complacent acceptance of a terrible update status quo, but I am saying we should bring back a little bit of the old outrage at carriers, manufacturers, and Google itself.

The many features in Android 9 Pie cohere into something that feels more polished than the last few versions of Android. There is a lot to like and fewer excuses than ever for updates not to come out for existing phones in a timely manner.

8 Verge Score

Good Stuff

  • Notification management still great
  • AI-based shortcuts
  • Better screenshots, text selection, and auto-rotation

Bad Stuff

  • Gesture navigation not as smooth as it could be
  • Some of the best features are Pixel-only
  • Android OEMs still need to get better at releasing updates

We've been living with the same three-button core navigation system in Android for several years now, but with Pie, Google is finally giving a gesture-based interface a shot. It may not be the most important new feature in the OS, but it's certainly the most prominent and the most divisive. Bear with me here because I'm going to overthink this, but I think it's worth it because it illuminates a key point about Google's design direction.

The new system replaces the back, home, and multitasking buttons with a singular home button, gestures, and other buttons that appear on an as-needed basis. In theory, it will make future Android phones more accessible to users who are used to the iPhone X's gesture system, and it also offers some benefits (swiping requires less accuracy than tapping). Overall, the new gesture system works, but it's conceptually complicated.

To see what I mean, here's a brief description of how gestures work: You swipe up once to get to an overview pane. The Overview pane (aka your recently-used apps) lets you swipe between apps or enable split screen with a hidden menu on the app's icon. On Pixel phones, you'll also get an AI-driven list of suggested apps and a search bar. Swipe up again, and you'll get to the app drawer with icons for all of your apps. You tap the home button to go home, or you can drag the home button to the right to quickly switch between apps in a screen that's similar to, but not identical with, the Overview screen. Along with all of this, the traditional Android back button will still show up from time to time next to the home button because Google hasn't yet developed a gesture for "back."

It's... a lot. I'm not against complication in principle when it comes to UX — I have faith in humanity's ability to learn — but there's no denying it takes some time to feel like you know your way around.

The funny thing is, I think the negative reaction isn't about how complicated gesture are. Instead, it's about how they feel. As I've written before, switching to a primarily gesture-based navigation system is a risky move for Google, because those systems only feel good if they... feel good. Any "jank" in the animation or weirdness in the physics of moving elements on the screen will make a user feel unmoored and unhappy.

The good news is that — at least on the Pixel 2 XL — Google finally got to a place where the animations work as they should, and the jank is gone. But the physics and ergonomics still feel a little off, especially if you're used to the system on the iPhone X. (After a rockier beta, animations were also fine on the Essential Phone with the final version.) Where the iPhone's gestures let you flow from one thing to the next with a single gesture, Android's feel a little more staccato.

As just one example, you theoretically have the option to do a long swipe up to get to the app drawer instead of a double swipe (once to the overview, once again to the drawer). But in practice you have to do a loooong, loooong swipe to get it to work, which you'll invariably get wrong, and the dock will give you a fussy little bounce in a futile attempt to indicate you should just double-swipe up.

I'm overthinking all this in part because I don't think Google thought it over enough. I would have jettisoned the long swipe and just encouraged people to double swipe. That would have the side effect of pushing people into the Overview screen more often, which would be a net good for Google. The app suggestions are very often exactly what I want and the swipe-tap motion to start a search is faster than any mobile search UX we've had, going on seven years (since, you guessed it, just typing on the physical keyboard of a webOS or BlackBerry phone).

But, of course, that enhanced Overview screen is a Google-exclusive feature. Other phones, like the Essential phone, don't have those Googley-bits at the bottom, they just have your app dock and no search bar.

With Pie, Google is leaving the buttons as the default navigation for current phones, and users will be able to switch back and forth from buttons to the gestures. Choices are nice but offering them instead of just going with what you think is best often reveals a lack of confidence. As you can tell, I share what I sense is Google's lack of confidence in the current system.

Despite all this belaboring, I do prefer the gestures to the buttons! It's a lot easier to just swipe up anywhere from the bottom of the phone, and I've used the copy-and-paste trick directly from the Overview screen a few times now. I just think they need a few more tweaks, and I suspect those will come in due time.

The new horizontal Overview / multitasking screen is the biggest visual change, but there are plenty of other nips and tucks around the interface. Nothing here will really feel alien to longtime Android users, it generally is just a bit more elegant.

Android still maintains its lead in usable, manageable notifications. They have a slightly cleaner layout, and the entire notification panel has rounded corners. There are still multiple priority levels, grouping, an overflow area, and no distinction between what's shown on the lock screen and notification panel. If you dismiss a notification from an app a lot, Android will eventually prompt you to just turn it off completely.

The quick settings panel up top has been simplified (some would say oversimplified), requiring you to long-press to access more settings instead of giving you an in-menu dropdown button. As it does with literally every revision, Google has also adjusted the main settings screen. There are colorful icons for settings, and it's more prominently adding suggested settings at the top more often than before. A system-wide dark theme is now an option for everybody whether you have a dark wallpaper or not.

Back to polarizing changes, though: the status bar has been rearranged to better accommodate phones with notches (which apparently is going to be damn near all of them not made by Samsung). The time has been shifted over to the left of the screen and the little notification icons that appear over there are capped at four, whether you have a notch or not. It was a necessary step given the hardware trend, but I'm hoping that eventually manufacturers will be able to report how much space their notch is taking so Android can display more icons if there's space for them.

Google has changed the volume button behavior a bit — they only adjust media volume now with a little on-screen pop-up that lets you toggle your ringer between vibrate, silent, and on. It's more predictable, and I think most people will prefer this behavior, but I'm an old person who actually adjusts ringer volume a lot, so it's less convenient for me.

The other little pop-up on the right side is the power menu, with options for restarting and taking a screenshot. I recommend hunting down the "lockdown option" in settings, which adds another button to that menu. Tap it, and your phone will require a passcode instead of letting biometrics unlock the phone. Honestly, that button should have been set to "on" by default.

I'm not sure what took so long, but Android finally has a magnifier when you're trying to move the cursor when selecting text. Another "finally" is screenshot markup. When you take a screenshot now, you'll have an option to crop it and draw on it before saving or sharing.

Last but not least, if you're the sort of person who leaves rotation lock on, Google will pop up a little button when you turn the phone to temporarily let you put it in landscape mode. Something about big phones has always caused them to be too aggressive at rotating the screen for me, so it's a nice feature. It can be annoying, though: most of the time you want to go 90 (ahem) to watch video, and video by default hides the main navigation buttons. It's a few extra taps to get back to portrait.

In my initial look at Android 9 Pie, I called it Google's "most ambitious update in years." I still think that's true, but unfortunately, right now, Android doesn't quite reach those ambitions. There are two key features that aren't shipping until later this fall: the so-called "Digital Wellbeing" dashboard and a feature called Slices.

Digital Wellbeing is available as a beta, and I'll wait until it's official to review it. But even in beta, it's useful. You can see how much time you're spending in apps, set limits, and turn on a great feature called "Wind down," which toggles on Do Not Disturb and sets the screen to monochrome. Honestly I wish there was a way to turn on Monochrome more easily anytime.

Do Not Disturb, by the way, has changed a little. It's much more aggressive at hiding notifications by default, down to not letting you see them at all unless you mess with the settings or turn DND off. The default is a little overbearing for my tastes, and I wish there was a better middle ground.

Slices are part of Google's initiative to bring more AI and machine learning to Android's interface. The idea is that the functions of an app can be "deconstructed" and spread out to other parts of the OS. So when you search for a thing you want to do, an app can show it's own interface or button directly in the search results. The commonly-cited example is hailing a car. We'll test it in the fall when it becomes available.

But there are other AI elements to Android that are available right now. Both battery life and screen brightness are automatically handled by machine learning that adjusts settings based on your usage. I can't really say how effective either are with any level of confidence, but anecdotally I do think I've been messing with screen brightness less often. AI also determines which icons appear at the bottom of the Overview screen, and it's crazy good — the app I want to open next is there at least half the time.

Finally there's "Actions," a feature that complements Slices and is available now. Where Slices will show buttons for app actions when you actively search for something, Actions puts those buttons directly in your app drawer. As with those icons in the Overview screen, Android tries to guess what you might want to do, only here it's a button that deep links into a part of the app. It might be sending a text or opening the podcasts app before you start your commute. They seem fine, but I'm not in the main app drawer often enough to make heavy use of them.

Android 9 Pie is a great update, and I wouldn't want to go back. I love that it's chock-full of ideas about how an operating system can be smarter, even though some of them (pardon the inevitable pun) don't feel quite fully baked.

I see a few trends beginning to come to fruition here. Through battery management and notification changes, Google is continuing its efforts to corral an ecosystem of bad-acting apps through a better-managed OS. The other big trend is one I've been talking about for a couple years now: moving towards making AI the new UI.

Two years ago at the Code conference, CEO Sundar Pichai told Walt Mossberg that Google intended to be more "opinionated" about its own phones, and the Googlification of Android on Pixel phones is stronger than ever now. The heavy emphasis on the Overview screen, Actions, and (eventually) Slices are all examples of Google trying to use its own AI chops to surface what you need instead of making you hunt through home screen folders and apps. It's been fascinating to compare Google's strategy to Apple's with iOS 12 — and will continue to be.

Of course, if we're bringing up iOS, we have to circle back to where we started: updates. Apple still trounces Android when it comes to getting phones updated to the latest OS. Last year, Google built the Treble infrastructure to make it easier for companies to push out these big OS updates faster. This year, I'd like to see more companies take advantage of it. Android users have more reason to hope than we have in a long time; the Essential Phone was the first non-Google phone I can remember that got an update the same day as Pixel phones. But that's just one phone out of hundreds (or more).

As happy as I am with all the individual features in Android 9 Pie, I'll be even happier if the Android ecosystem gets its act together and releases it.

Let's block ads! (Why?)

ARM challenges Intel in PCs with Deimos and Hercules chips

Posted: 16 Aug 2018 06:02 AM PDT

With PC makers like Asus and HP beginning to design laptops and tablets around ARM chips, ARM itself has decided to emerge from the shadows and unroll its roadmap to challenge Intel through at least 2020.

ARM’s now-public roadmap represents its first processors that are designed for the PC space. ARM, taking aim at the dominant player, claims its chips will equal and potentially even surpass Intel's in single-threaded performance.

ARM is unveiling two new chip architectures: Deimos, a 7nm architecture to debut in 2019, and Hercules, a 5nm design for 2020. There's a catch, of course: Many Windows apps aren’t natively written for the ARM instruction set, forcing them to pay a performance penalty via emulation. Comparing itself to Intel is a brightly-colored signpost that ARM remains committed to the PC market, however.

What this means for you: ARM-powered PCs like the Asus NovaGo offer game-changing battery life—but the performance suffers, for two reasons: One, because the computing power of ARM’s cores has lagged behind those of the Intel Core family; and two, because any apps that the ARM chip can’t process natively have to be emulated. ARM can’t do much about Microsoft’s development path, but it can increase its own performance. Finally, if you were concerned that ARM PCs will be a flash in the pan, the answer is no, apparently not.

pcmark work asus novago Mark Hachman / IDG

Where ARM is starting from: this PCMark Work score from our Asus NovaGo review (powered by an ARM chip) put the NovaGo at the bottom of the barrell in terms of performance.

ARM, playing catchup to Intel’s Core

ARM, whose chips already dominate the smartphone space, doesn’t supply chips directly. The company is what’s known as an intellectual property (IP) provider, and a “soft” one at that—it designs and licenses chips to companies like Qualcomm and Samsung, who are then free to modify and eventually manufacture them for smartphones and PCs. That practice will continue, according to Ian Smythe, responsible for ecosystem marketing for the client businesses for ARM.

This structure can make who-supplies-what a bit confusing. What you need to know is that ARM continues to roll out new designs on essentially an annual basis. Qualcomm—which is, for now, the exclusive supplier of ARM-based chips for PCs—follows suit with its own modified Snapdragon chips, though a bit later. PC vendors then have to take those chips and roll them into PCs, which takes yet more time.

arm perf roadmap ARM

ARM's roadmap paints an aggressive comparison between single-threaded performance between its own chips and Intel's Core architecture.

The Asus NovaGo incorporated the Snapdragon 835, a 2017 Qualcomm chip whose Kryo 280 compute engines were based on 2016’s ARM Cortex-A72 core. Qualcomm then promised that the Snapdragon 845, a 2017 core based on ARM’s 2017 10nm Cortex-A75 architecture, would follow. But the 845 sort of vanished, replaced by the Snapdragon 850 at this year’s Computex show as Qualcomm’s official PC processor in waiting. Both the Snapdragon 850 and 845 used the same A75 ARM architecture, but at different clock speeds.

In the meantime, ARM announced its next chip, the Cortex-A76, this past May, promising a 35-percent performance improvement. (Qualcomm has yet to announce a Snapdragon based upon the A76 architecture.) Smythe also said that ARM's partners will indeed manufacture A76 chips on 7nm processes this year, maintaining a steady pace of manufacturing improvements, while Intel has seemingly stalled.

Asus NovaGo snapdragon sticker Mark Hachman / IDG

Could ARM's partners be the next Intel Inside? 

What this all boils down to is that the Asus NovaGo, criticized for its poor performance, is based on a two-year-old ARM architecture that’s steadily improving. ARM’s Smythe declined to comment on whether existing customers like Asus and HP would continue buying ARM chips. “But they are hopefully excited by the performance and power values that we’ve described today based on our A76 and beyond,” he added.

Can ARM actually lead in compute performance?

According to ARM’s roadmap, the A76 (expected to run at about 3GHz) represents a sharp jump in performance. Smythe said that ARM ran SPECINT2006, a single-threaded benchmark, and compared the scores through the Core i5-7300U to its own performance estimates. Granted, it’s only a single benchmark, measuring single-threaded performance up through Intel’s seventh-generation Kaby Lake chip. (Intel is currently shipping Kaby Lake-R chips, and will move on to Whiskey Lake by the end of the year.) And it excludes the Core’s turbo mode, Smythe added.

Single-threaded performance remains a poplar metric, if only because programmers haven't done a good job of taking advantage of the parallelism inherent inside chip architectures like Core and ARM. It's here that ARM has traditionally lagged far behind. 

For example, Anandtech published a 2016 comparison of Cavium’s ARM-based server chip with a comparable Intel Xeon chip—well before ARM was considered a PC processor. Then, ARM processors generated 25 percent or so of a Xeon’s performance. Parity—or even above—would be an impressive leap.

“If we look at projected performance for 3GHz A76, it sits right over at the high end of the performance range of the Core i5,” Smythe said.

arm vs core clock ARM

Comparing the A76 to an older Intel chip isn't fair, though it's a starting point.

It also discounts ARM’s inherent power advantage. ARM builds its processors in what it calls a “big.LITTLE” configuration, pairing 8 or so A76 compute engines with power-optimized A55 CPUs to take over in idle mode. That helps the A76 consume under 5 watts, while the Intel Core i5-7300U consumes about 15 watts, trimming battery life.

What this boils down to, Smythe said, is that ARM will begin talking more loudly about its place in the PC space. “We can now demonstrate where we are exactly in terms of performance,” Smythe said.  “So are we going to be more aggressive? Yes, I think that’s fair.”

To comment on this article and other PCWorld content, visit our Facebook page or our Twitter feed.

Let's block ads! (Why?)

Google rebrands cloud storage services as Google One with cheaper plans, extra benefits

Posted: 16 Aug 2018 04:56 AM PDT

  Google's rebranding of its cloud storage service from Google Drive to Google One has taken a major leap forward, with the tech giant now allowing anyone in the United States to sign up for the plan, not just users already paying for online storage capacity.


Announced earlier this year, Google One is effectively a renaming of its cloud storage service, one that encompasses many other Google services that have online storage components. Users can pay a monthly fee to Google One for storage, which can be used with Google Drive, Gmail, and Google Photos, expanding free allocations and providing more space for files.

Since its announcement, Google has been slowly converting its consumer cloud storage customers over to Google One, but now it has been opened up to anyone in the United States. It is unclear when it will be available to users in other countries, but Google is offering to notify potential users when it opens up in their market.

The Google One plans start with 100 gigabytes of storage for $1.99, rising to $2.99 for 200 gigabytes and $9.99 for 2 terabytes, while higher capacities will still be available, but under previous pricing. The new options may be seen as better value to consumers wanting more storage, as 1 terabyte previously cost $9.99 per month, but the 100 gigabyte option could also be acquired for $11.99 per year before the rebranding effort.

Google is also attempting to make the subscription more useful for families, with customers able to share their plan with up to five other people under one bill. A number of other benefits are also offered, including Google Play credits and hotel deals found on Google Search, with offers from Google Express and Google Store expected in the coming months.

Support is also touted as a big feature of Google One, with users being offered access to a "team of Google experts" to answer their queries.

The rebranding to Google One and the refinement of the pricing brings Google's service close to Apple's iCloud in terms of value. While iCloud has the same prices for the 200 gigabyte and 2 terabyte options, it offers a lower 50 gigabyte tier for $0.99 per month.

Google does still have Apple beat in how much it provides to users for free, offering 15 gigabytes compared to iCloud's 5 gigabyte allocation. Apple also doesn't offer extra bonuses for iCloud subscribers, which could make Google One more attractive to those not embedded in the Apple ecosystem. </span>

Let's block ads! (Why?)

This post have 0 komentar


EmoticonEmoticon

Next article Next Post
Previous article Previous Post