The Windows RT Review
by Vivek Gowri & Anand Lal Shimpi on October 25, 2012 12:00 PM EST- Posted in
- Windows RT
- Operating Systems
- Microsoft
- Mobile
- Windows 8
- Tablets
Internet Explorer 10
There are two distinct versions of Internet Explorer 10 - one that runs in the desktop and looks very similar to the IE9 and IE10 experiences that we’ve been used to on the Windows 7/Windows 8 desktop, and another that runs within the framework of Modern UI and looks like an upscaled version of IE9 Mobile from Windows Phone 7.5 Mango. Both use the same rendering engine and perform identically, so we’ll keep that in mind when looking at performance numbers. Other than the user interfaces, there’s relatively little difference between the two browsers, though you will need to open the desktop IE10 to change browser settings - there's no way to do so from within Modern UI.
Gestures in the Modern UI version of IE are very important - swiping from the top or bottom brings up the URL bar at the bottom and the thumbnail list of open tabs at the top. It looks great, and keeps the webpage completely clean, but I don’t think the tabbed browsing implementation here is that great - I’d rather see something like Chrome or Safari with the list of open tabs always displayed at the top of the window. Changing tabs requires an extra step (swipe, then select), which isn’t ideal for changing between tabs quickly. If you deal with a lot of open tabs, or have some need to flip between two specific tabs repeatedly, this will get annoying relatively swiftly. It still looks great, and for touch-based browsing it works quite well other than the concerns over tabbed browsing. The “fancy” version of IE10, as Anand called it, strikes me as a very idealistic design philosophy that doesn’t necessarily work as well as a less elegant interface might.
One feature that works very well but could use some performance tuning is the back/forward touch navigation for Modern UI's IE10. To go back a page just swipe from left to right (or right to left to go forward). It's incredibly intuitive. My only complaint? Although going back immediately shows you the previous page, you have to wait for the page to actually re-render before it's usable (which on present day ARM hardware isn't exactly fast). I suspect this is something that will become far more useful over time with faster SoCs.
The desktop IE10 is just desktop IE, you’ve seen this before. Nothing that new, the UI looks pretty similar to IE9 (which is the first version of IE I can remember enjoying, though it will never replace Chrome in my heart) and it functions similarly as well. Obviously, there’s an updated rendering engine, but other than that and a new scrollbar design to match the rest of the Modern UI visual style, there’s not much to see here.
The most interesting part, actually, is in the settings - you can set links to default open in Fancy IE10, the desktop IE10, or let IE decide which to use on a contextual basis. You can also have the IE10 tile in Modern UI to open the desktop IE10, so you can basically ignore the new IE10 UI entirely should you want to do so. Now, I suspect this option exists primarily for desktop and notebook users of Windows 8, who don’t want to deal with using the Fancy IE10 interface with a mouse input. But it’s there for Windows RT users who want to use only the desktop IE10 UI. Personally, I wouldn’t want to use the classic interface in touchscreen devices - UI elements are damn small at this pixel density, making it more difficult to hit the right buttons than it needs to be. There’s a reason the Modern UI variant of IE10 exists.
Performance
From a performance standpoint, we’ve just brought over the same three benchmarks that were used in the VivoTab RT and Surface RT reviews - SunSpider, RIABench, and Mozilla's Kraken. Honestly, I think SunSpider scores are becoming irrelevant - as one of our commenters said, the art of getting better SunSpider scores has started becoming a race to see who can cheat on the test most. Cheating in this case would be considered optimizing browsers for this specific test, and they’re not wrong. While IE10 has great, great SunSpider scores, it’s far behind in the other two JS benchmarks compared to similar hardware running on other platforms - look at the Surface or VivoTab RT compared to the international Tegra 3 version of the HTC One X. And I’m almost certain that if we had a Nexus 7 on hand to run those tests on, the differences would be larger still.
The HTML5 Test score 320 (plus 6 bonus points) is better than most tablet browsers from a year ago, but lags behind the latest mobile versions of Safari (386+9) and Chrome (390+11). Our Windows 8 Consumer Preview article showed that IE10 had made some pretty significant strides from IE9 (which was leaps and bounds better than 8, and then some) but JS performance is still behind the curve for now. The experience doesn’t suffer from that though - it’s still a very capable modern browser that won’t leave you disappointed.
Facebook
Twitter
RSS
233 Comments
View All Comments
Dorek - Friday, November 2, 2012 - link
AFAIK, task manager tray doesn't even have thumbnails. Come on, it's not comparable to what Windows 8 and Windows RT do. Have you used them?yyrkoon - Sunday, October 28, 2012 - link
The software issue, will not be much of an issue at all.Microsoft will just write another abstraction layer ( HAL ) for ARM. Which is most likely what they've done here.
Passed that. IF any abstraction layers need to be written for the COM model, or .NET libraries. You can rest assured Microsoft will cover that as well. If it hasnt been covered already. Which Im betting it has.
Again, this is nothing like WinCE.This is a real OS, for real end users, with a different HAL.
The biggest issue as I've stated in previous posts. Is not whether it will be easy to write applications for Windows RT. But whether software developers deem it prudent to spend thousands of dollars on Microsoft development tools( not absolutely necessary ). THEN pay Microsoft a forced 30% distribution fee. Through their app store.
For me personally. The 30% fee alone is enough for me to jump ship. And I really enjoy writing apps for Windows. We'll see how it all works out in the end.
karocage - Friday, October 26, 2012 - link
This comment from the article seemed odd to me:"This isn’t like Windows Phone, where we need to see whether the platform will get any market traction before predicting the growth of the app marketplace."
Market traction and app store growth don't seem to have much if any relationship with Windows Phone. WP7 hit 100,000 apps faster than either iOS or Android and has over 125,000 now. If anything, that would seem to indicate that MS really does have a lot of pull with developers independently of market share. So while I agree that the sheer number of devices will be a big draw, I'd argue that's just the cherry on top that guarantees we're going to see a ton of support really quickly for RT/Metro apps.
Obviously buying a WinRT tablet today will be pretty limiting in terms of 3rd party software, but experience from WP7 suggests this will be a very short-lived downside of the platform.
Dorek - Friday, November 2, 2012 - link
Good point. I've been on Windows Phone 7.5 for a year, and have never had the problem of not being able to find an app I want. (With the exception of Pandora, the assholes. Can't even use third-party ones because they block them. But I have a Zune--I mean, Xbox Music Pass anyway. So screw Pandora.)andypost - Friday, October 26, 2012 - link
Anandtech writes the best articles. period.Thanks to Vivek Gowri and Anand Lal Shimpi for a great article.
Visual - Friday, October 26, 2012 - link
You say it is impossible to make desktop apps, but...I read that Win RT still has PowerShell. A lot of PowerShell scripts on x86 just invoke various functions in third-party ActiveX controls, which are just x86 executable binary libraries, and that will not work on RT... but maybe at least the core MS objects still work? For example, if Windows Forms or Windows Presentation Foundation still works, then PowerShell scripts can create desktop windows with various GUI elements.
And since there is an IE, maybe HTA applications work as well?
Do you have some scripting geek working at AT that could test these theories out?
On another topic, I am very surprised you say there are x86 executables on your install. That is just wasted space on a device that doesn't get a large drive to begin with anyway. MS would get in serious trouble if it were true... Are you sure?
And lastly, I am curious if a Win RT device is supposed to be able to get plugged to a "real" Windows computer over USB for charging or for as a mass storage drive, for transfering files... If it is, then coudn't you also plug one RT device to another RT device? Then which one will become the "host", and which one will become the peripheral?
Dorek - Friday, November 2, 2012 - link
"And lastly, I am curious if a Win RT device is supposed to be able to get plugged to a "real" Windows computer over USB for charging or for as a mass storage drive, for transfering files... If it is, then coudn't you also plug one RT device to another RT device? Then which one will become the "host", and which one will become the peripheral?"Dude...that's deep.
(That's actually a really good question. Surface can't be charged over USB, but I don't know about other ARM devices.)
tommo123 - Friday, October 26, 2012 - link
i mean if someone writes an app and gives it away. there's no sideloading so if has to go through the store right?will microsoft host it for free? if not then they're effectively killing off free software for windows are they not? at least win rt which is eventually (how they'd like) desktop windows to go too.
tipoo - Friday, October 26, 2012 - link
Yes they'll host it for free. And there are many free apps already, there was a influx of apps on launch day and most reviews were done before that. They only take a percentage of the cuts from paid apps, nothing from free.Visual - Friday, October 26, 2012 - link
I forgot one question in my previous post.How is the browser performance with many many tabs open at once? Like 20, or even 100?
I sometimes use tabs in a bit weird way, ctrl+clicking possibly interesting links in some large index page to open them in a background tab for later reviewing it in more detail. Even without Flash, with IE9 starting a new process for each tab by default and a few animated gif ads and various JS scripts on each tab, it really kills Atom Netbooks and even CULV Core 2. It's probably more of a memory usage thing than a CPU thing actually... But anyway, how could that go on these new ARM devices?