42 comments

[ 3.3 ms ] story [ 87.8 ms ] thread
"One OS to rule them all"

Nice try, Microsoft, misleading people into thinking that. But it's more like:

1) Windows 8 for x86 desktop systems

2) Windows 8 for ARM (no native apps, and quite different)

3) Windows Phone for mobiles (based on Windows CE)

My question is why didn't they just make WP7 for tablets? I know Win8 wasn't ready, but from some videos I saw on TIMN of Win8 running on ARM tablets, it was clear that the dual core ones could barely handle it, and only the quad core Kal-el had more acceptable performance. I'm sure WP7 would've flied on those. Why did they have to port that whole 8 GB OS to ARM when it will have no legacy apps anyway?

> 2) Windows 8 for ARM (no native apps, and quite different)

I think the article might of worded it badly, but their will be native ARM apps for W8 on ARM, but not x86 emulation. It pretty much is the same as Windows NT back in the day.

> might have

> but there will

FTFY. Sorry.

thanks, I guess. I should have used correct grammar while commenting on an article being poorly written. It is a shame that you have received down votes for it (I did up vote).
I'm a bit confused. People expected x86 emulation on ARM architectures? That sounds a bit backwards to me.
Yes, this article is way off target. Who exactly expected ARM Windows to run x86 applications? Show me one reference. Anyone?
When Intel trumpeted the fact that ARM Windows 8 machines wouldn't run legacy apps, MS reps called it "inaccurate and misleading". From the article: http://www.theregister.co.uk/2011/05/19/microsoft_contradict....
Well, since a lot of Windows software runs on .net framework's CLR (VM with a bytecode interpreter not unlike the Java VM), a lot of legacy apps actually should run just fine on ARM.
I honestly expected that they were going to try at least for the initial release. Apple did release Rosetta when they moved to Intel, which allowed PowerPC-based applications to run on x86 Macs.

I'm not sure how much use that got, though, considering that Rosetta had a lot of limitations, and fairly terrible performance in a lot of areas. That's probably the main reason that MS didn't bother trying to support it.

Rosetta saw plenty of use just from Adobe apps. It took them about a year and half IIRC to release an x86 version of Photoshop.
edit: and didn't Office:Mac take forever too?
yeah, office 2008 was the first native x86 version, IIR (and I swear it takes longer to start up then the PPC version of 2004 did under Rosetta, which is just ridiculous).
Rosetta worked because the switch to intel processors gave Apple such a performance boost compared to their PPC systems, that existing PPC applications worked with about the same performance as they used to on PPC native.

The same probably wouldn't be true for i386 -> ARM.

So apparently when they say "ONE OS TO RULE THEM ALL" what they really mean is "ONE BRAND NAME TO RULE THEM ALL with a number of non-cross-compatible OS's which have the that brand name and a similar UI"
I wonder if they will go the other way though and emulate ARM on x86.
Why would they? Only reason you would need emulation is for legacy apps, which Win8 will have none of from ARM.
Development, sure it could run natively like iOS does(they just reroute the system calls), but it means you have to do more on device testing since the translation layer isn't bug for bug compatible with the device os.
So, there was someone who expected that x86 win32 apps will run on ARM? ^_^

So, that rumors about miserable average level of intelligence in US aren't just rumors? ^_^

Why aren't they adding support for fat binaries (or something similar) like on MacOSX? So that developers can indeed release one single version of their app which runs on all the different architectures.
Notice there is no iOS/OSX fat binary - because the only approved apps for iOS are all from the App Store.
If you write your app on HTML/JavaScript or (presumably) .NET, it should run on both platforms.

If you are deploying native code, I would think that creating your own "fat binary" would not be too hard to do. Either install the right binaries with a MSI file or at runtime have the startup program be a small managed app that starts the right native app for the platform.

x86 apps on ARM is a distraction. The question is whether Win32 apps will run on ARM; MS has given conflicting information on this topic.

http://thisismynext.com/2011/09/16/windows-8-arm-tablets-leg...

It looks to me like it's WinRT or nothing for Metro apps in general.
Could you explain what you mean? Isn't "Desktop mode" Win32?
Yes. If Win32 is not available on ARM, maybe ARM systems won't have desktop mode at all.
I would highly doubt that.

'Desktop' can be seen as a panel on the start metro UI[0]. Remember, not all tablets will be ARM systems.. and I could see a market for ARM desktops too.

IIRC MS showed the desktop running on their very first demonstrations of v. early win8 on ARM, too.

0. http://www.engadget.com/2011/09/13/windows-8-for-tablets-han... @video.

So in tablets, they're going to be in the same position as Windows Phone 7 - a great OS with a distinct app disadvantage and little consumer awareness. This only further raises the question of how much appetite device makers will have for a $75 (ballpark) software license, especially as these devices start to average closer to $400.
My guess is that for ARM devices they'll use a different licensing scheme with a price tag closer to windows phone 7 of $5-10.
I doubt it. I think one of the main reasons they chose to use the "real" Windows for the ARM tablet version instead of WP7, was because they wanted to charge several times the price for it. So expect them to charge at least $30-$40 for the ARM version, even though it basically works the same as the WP7 version would've worked on tablets, with the necessary modifications. In fact, it would've probably had much better performance, since it's much lighter.
I think it puts them at a competitive disadvantage to charge so much for a more limited version of windows. I think they chose to use Windows 8 to create confusion that legacy windows applications will work on tablets.
Personally, i think microsoft were just scared of losing the tablet platform to apple and google. Desktops, or x86, very possibly aren't going to be common in the forseeable future. Windows is their clear competitive advantage to leverage.

Many full desktop apps will be compilable to run on ARM, and lets not forget MS's own software; full office and internet explorer put windows8 ahead of the competition re: serious work already. (Besides windows phone has already bombed on phones, no reason to suspect it to fair any better on tablets).

Going forward it will be much easier to develop for both platforms. This is a very similar approach to the XBOX; there is an easy way to make software for both platforms.

I'd be really really suprised if the large scale PC producers were sending MS anywhere near $75 for each PC with windows on it. I'd say the profit on a PC not accounting for the licence fee would very likely often be less than $75. I would think there would have been a much greater push for linux or no-OS installed sales if it were anything in that ballpark.
If we estimate around 80MM PCs sold in a quarter, MS generates about $60 Windows revenue for every PC sold, but their costs are about $22 (from Q411 results Rev $4.7B, Profit $2.9B). It will be hard for them to the cut license fee unless they can significantly increase unit sales.

This is probably why they didn't push WP7 on tablets.

For almost all the OEMs, the average margin on a PC is single digits.

Ah, thanks for the specifics. This only underscores the existential threat that a decent quality $300 Android tablet poses. Microsoft would need a staggering advantage to even sell Windows 8 at cost ($22) to a manufacturer at that level.
Can we get some citations on these numbers please? ..we aren't even in Q4.11 yet?

Another key factor to bear in mind is that, at least today, tablets are an additional computing device, and don't replace any existing PC.

Fiscal years don't always match calendar years. Microsoft's Q4 ends on June 30th.

http://www.microsoft.com/investor/EarningsAndFinancials/Earn...

I don't think you can say that tablets are bought in addition to PCs, as we've seen PC sales flatline and contract in the U.S. and Europe since the iPad was released. Nobody really knows what the long-term impact will be.

tablets are an additional computing device

Why? As of iOS5, none of the tablets require a desktop or laptop. I'd venture to say that %90 of people use their home computer for communication (social networking, e-mail, skype) and leisure (games, web-browsing). Why should someone buy something heavy and expensive when a sleek, inexpensive (and dockable) tablet meets all of those needs?

I see this playing out much like the Apple OS 9 -> Cocoa / Carbon migration. Smaller, more nimble developers will convert to or create new apps on the Metro stack, while established developers will stick with x86 only apps with a smattering of "new platform experiments" (see Adobe's behavior on OS X / iPad).
Microsoft will incentivize developers through various means to help move apps to the new WinRT platform fairly quickly, thus help seed the ecosystem with ARM compatible apps. This isn't the same Windows Phone 7 apps chicken and egg problem. Microsoft will be pushing millions of Windows 8 computers into peoples homes sometime next year, which will help the seeding process. When this happens, I suspect we will see Microsoft provide preferred Windows Store placement for Metro apps, and a "Developed for Metro" strategy that makes the legacy x86 app experience seem antiquated and clunky. This upgrade sales opportunity will incentivize developers to push out new versions, which will help seed the ARM compatible apps ecosystem.
I've got to side with Microsoft on this one. I see no evidence that any of Microsoft's statements were misleading, not to mention that I think this break in backwards compatibility will only be a good thing for performance on ARM devices.
M$ is trying very messily to 'be the next Android'. So, what's new?