X
Tech

Microsoft's Metro proves the PC is dead

The Metro UI and the WinRT APIs will signal the end of the traditional Wintel platform and usher in a completely new generation of Personal Computers that will have little resemblance to their forebears.
Written by Jason Perlow, Senior Contributing Writer
davidburela.png

The article which I published last week, "Post-PC: Why Intel Can No Longer Live In Denial"which was an expansion of the Post-PC Great Debate kickoff between myself and Zack Whittaker apparently struck a sour chord with a lot of folks. 

I'm going to address those themes and tell you what comes next.

First, a few folks were under the impression that I have an inherent bias because I work for IBM, which competes with Intel-based server systems. Before I go any further with the main subject of this post, I'd like to state that this is most definitely not the case. In fact, most of the time, I really try to go out of my way not to write about IBM at all. IBM also has a corporate blogging policy that requires me to keep my nose clean.

But the IBM mentions are unavoidable when talking about big iron architectures. There just aren't many of them. So when I do mention IBM in an article (which is inevitably bound to happen) I write a disclaimer at the end of each blog post that my opinions are strictly my own and not IBM's.

I also have a separate disclosure page on ZDNet which states quite clearly that I'm employed by IBM, just so there are absolutely no ambiguities here.

To me, it's obvious that the x86 is now going to have to compete with Oracle's/Fujitsu's SPARC enterprise servers, as well as IBM's POWER enterprise servers and System z Mainframes for running Cloud workloads. The industry has heavily consolidated and now only Oracle/Fujitsu and IBM make "Big Iron" processors.

They are the only companies that can afford the R&D to do so. If I could list more vendors that make Big Iron to compete with Intel's high-end products I would.

Big Iron just don't exist outside of that small group of vendors, period.

On the personal computing front, I think ARM is the platform that will replace Wintel on PCs. ARM also happens to compete heavily with IBM in the embedded chip market. More on that later.

Now back to our regularly scheduled programming and my theory that the x86 PC will be pushing up daisies in 10 years.

I'd like to direct you to the header graphic of this article, with the two big green and blue boxes that contain smaller green and blue boxes. This graphic comes straight from the Microsoft BUILD conference in Anaheim last week.

If you're not a software developer, this graphic (and the various versions that have been shown in Mary-Jo's article) might not make a whole lot of sense to you. But let me simplify it for the layperson so that the concepts are easier to understand.

Everything in the green is the "New" Windows stuff that will release with Windows 8, which includes programs and sub-systems which run under the new "Metro" user interface, or the Windows Runtime, also known as WinRT.

If you want a good look at Metro and WinRT up close, watch my Windows 8 Developer Preview Tour video.

Everything in the "Blue" on that diagram is the old-school Windows we are all currently using. This represents almost 20 years worth of legacy Microsoft technology that originated with Windows NT 3.1 from way back in 1992, namely the Win32 Application Programming Interfaces (APIs) and also successive iterations of the  .NET Frameworkwhich were introduced in 2002.

I won't even get into 3rd-party programming environments such as Java and Adobe Flash/AIR, because it will get far too complicated.

Software developers use these APIs in order to write major desktop applications such as Office or even PC games like World of Warcraft, and they have over time been enhanced and grown to tens of thousands of API calls in total.

While both the "Green" side and the "Blue" side share common programming languages, such as C, C++, C#, Visual Basic, HTML and Javascript, the APIs which you use to write the applications to -- the frameworks, the function calls, all of the things which make up a complex software program such as Microsoft Word or Excel are completely different.

What this means is if you've written a complex application on the "Blue" side, you will need to completely re-write substantial portions of your application if you want to move to the "Green" side.

Microsoft has not yet said whether or not they will provide rapid porting tools from "Blue" to "Green", or even what their own internal time frame is for re-writing their own major desktop applications like Office are.

Presumably, this is a process that is expected to take several years.

What they have said, however, is that the big "Blue" box for all practical purposes will not exist on the ARM platform at some point in the future. There will only be "Green", aka WinRT.

Despite my long-time friend and colleague Mary-Jo Foley's claims to the contrary, there is no way, no how, that Microsoft is porting the entire "Blue" stack to the ARM architecture.

I could see porting small parts of it, perhaps, such as the Windows 7 shell environment, in order to maintain a consistent UI across the two architectures, and certain subsystems to support DirectX games, and providing a way for RemoteFX seamless RDP apps to run on a desktop on an ARM tablet, but not the whole thing.

There's just far too much legacy baggage involved.

Certainly, the most recent video demo of a prototype nVidia Kal-El based Windows 8 ARM tablet previewed on Joshua Topolsky's This is My Next blog does show the Windows desktop shell. But that doesn't mean the entire Win32 and .NET stack is going to be available to software developers natively on ARM.

The effort to port all of that over would be a Herculean task with very little payoff, and would defeat the purpose of bringing developers into the new WinRT "Green" environment in the first place.

The most logical way to provide legacy compatibility would be to provide seamless RDSH/RemoteFX client connectivity on ARM, with all of the key productivity apps hosted in Microsoft's Cloud (Office 365/Azure) or in private datacenters.

I've discussed this already in my article on Windows Server 8, which is positioned very strongly as a Cloud-enabled OS.

So if "Blue" doesn't exist on ARM tablets, ARM smartphones, ARM thin and smart clients, and Microsoft wants everyone to eventually move all their apps over to "Green", what exactly is the point of keeping x86 around on the desktop?

There's no justification for it whatsoever.

In my May 2011 article Project Blade Runner: The Personal Computer of 2019 I discussed a completely theoretical "PC" composed of next-generation parts, and talked about what the operating system on this theoretical personal computer could look like, based on current trends in virtualization and Cloud Computing.

Now that I've seen Microsoft's next-generation Windows desktop systems architecture, it's become very apparent that Metro fits quite well into the Post-PC vision that I described earlier. It would be entirely possible for a "PC" to be built entirely without Intel legacy architecture, based on the new "Green" foundation.

Within several years, the Wintel architecture on the desktop is going to cease to exist. Microsoft has not issued a timeframe for when this transition is going to occur, but I suspect it will happen within two consecutive versions of the OS. That's certainly within the scope of ten years or less.

Those who doubt that ARM will be capable of running powerful content creation apps will discount this notion completely. And certainly if you look at the platform today, it doesn't have the 64-bit architecture that is needed to pull off the rich media types of stuff you see on high-end workstations or even on gaming PCs.

I'm not going to sugar coat this. It will take several years for ARM to get 64-bit instructions and 64-bit memory with heavy core parallelism into mainstream chips to make them perform comparably to what exists in x86 desktop PCs today.

So initially, ARM will be suited best for Tablets and Thin Clients, where legacy Win32 and .NET apps can be accessed on VDI residing in the datacenter on Windows 8 Server using RDSH (RDP) and RemoteFX.

However, to quote Master Yoda, "There is another."

Microsoft already has a powerful hardware platform from which to draw on -- it's called the XBOX 360.

Wait, what? The XBOX 360 is a game console!

Well, not if you look at the basic guts and fundamental architecture of the machine. The XBOX 360 is in fact a PowerPC-based computer that has been tailored to run demanding games and runs a subset of the Win32 APIs.

The current version, the model S, includes up to a 320GB hard disk drive and 4GB of memory. The CPU is a tri-core custom chip called the Xenon, designed by IBM and manufactured by Globalfoundries.

There is certainly no reason why Microsoft could not take the WinRT "Green" layer, the Metro UI, a PowerPC port of Windows Kernel Services (perhaps using "Minwin") and throw it on top of a new 8-core, 16 hyperthread 45-nanometer Xenon chip with 8GB of RAM, a high-performance SSD like a MacBook Air, an improved GPU inside a small footprint casing and call this a Windows Metro Personal Computer.

Bundle this with the XBOX 360's existing developer environment for gaming, with a Windows App Store for native "Green" software, and you got yourself one hell of a product. In the future, once the chip platform reaches maturity, Microsoft could do the same with an ARM -- aka something along the lines of the "Blade Runner" that Scott and I have devised.

No Intel required. Why, it sounds almost... Mac-Like.

Crazy, I know. But certainly not outside the realm of possibility when you consider that when everything legacy is thrown in the garbage, the opportunity to re-design the Personal Computer as we know it and start completely fresh is extremely compelling.

Will Microsoft's Metro and WinRT radically transform the "PC" as we understand it today? Talk Back and Let Me Know.

Disclaimer: My Full-Time Employer is IBM. I write as a freelancer for ZDNet. The postings and opinions on this blog are my own and don’t necessarily represent IBM’s positions, strategies or opinions.

Editorial standards