Skip to main content

A closer look at the Surface tablet and Windows RT: Why Microsoft should never have dropped Intel

Over the past week, I’ve had the fortune to play with both Microsoft’s Surface 2 (which we'll have a review of very soon) and the Asus T100 Transformer Book. These are very similar devices – convertible laptops with detachable keyboards – except for one big and fundamentally life-altering difference: Where the Surface 2 is powered by Nvidia’s ARM-based Tegra 4 SoC, the Transformer Book has Intel’s x86 Bay Trail under the hood.

As a result, while the Surface 2 runs Windows RT, the T100 runs full Windows 8.1. Yes, every program and game that you use on your Windows desktop PC also works on the T100. Steam works on the T100. Team Fortress 2 works on the T100. Photoshop works (surprisingly well!) on the T100. Let that sink in for a moment, and then read on.

If it wasn’t enough to have a tablet-cum-laptop that can run the entire library of x86 games and programs, get this: The T100 boasts the same or better battery life than the Surface 2, performance is about equal (better CPU performance; worse GPU), and you get the same amount of storage and microSD card expansion. The kicker, though, is that the 32GB Asus T100 with keyboard dock is around £350 – the comparable Surface 2 with Type Cover is £470 (the disparity is more in the States, with the pricing being $350 and $580 respectively). Oh, I forgot to mention, the T100 comes with Office 2013 Home, too, and unlimited Asus cloud storage.

In short, the only real thing going for the Surface 2 is its 1920 x 1080 display. The T100’s 1366 x 768 display is still just fine for 10in tablet, but it’s definitely not as sharp as Microsoft’s unit. Still, when you’re talking about a £120 or $230 price difference, Asus had to cut some corners – and to be honest, while the T100 definitely feels cheaper than the Surface 2, it doesn’t actually feel cheap or tacky by any means.

Of course, this raises a rather important question: Why would you ever buy the Surface 2, or indeed any other ARM-powered Windows RT tablet? If Intel has finally closed the performance and battery life gap between x86 and ARM, why would you ever intentionally opt for a crippled ARM-powered tablet?

Another interesting comparison to make is the Surface Pro 2, with its fourth-gen (Haswell) Core i5 x86 CPU. Don’t get me wrong, the Core i5-4200U in the Surface Pro 2 is obviously faster than the Atom Z3740 in the Asus T100 – but is it fast enough to make up for the £480 (or $680 Stateside) price difference? In most workloads, the Haswell chip is between two and three times faster than Bay Trail. In some heavily threaded integer workloads, the quad-core Bay Trail Z3740 is very nearly as fast as the dual-core Haswell i5-4200U. There are other differences, of course – Haswell’s GPU is much faster than Bay Trail, and the Pro 2 has better expansion ports – but at the same time, the Pro 2 is much heavier than the T100 and really quite hard to use with one hand.

In real-world use, the main difference is that the Surface Pro 2 is capable of playing some modern-ish games at low resolution, while the T100 is mostly limited to 2D and indie games. Intensive apps like Photoshop will be faster on Haswell than Bay Trail, but unless you’re manipulating huge images it probably won’t matter; likewise, unless you’re juggling a bunch of Office docs, Bay Trail will be fast enough. And just to remind you, the price gap between the two systems is almost £500.

Bay Trail über alles

Over the last couple of years I have spent thousands of words trying to get my head around one of technology’s weirder quandaries: Why did Microsoft develop the ARM port of Windows 8 (now called Windows RT) in the first place? Less succinctly, why did Microsoft spend hundreds of millions of dollars, and millions of development hours, building and marketing a platform that has almost no redeeming features?

The thing is, it’s not like Microsoft didn’t know about Bay Trail. Bay Trail, or at least the out-of-order Silvermont CPU core, has been on Intel’s public roadmap for at least two years. Microsoft, being one of Intel’s closest partners, probably knew about it long before that. Even so, Microsoft made the active decision to forsake the Wintel alliance that had been the cornerstone of its dominance for 20 years, instead penning deals with ARM chipmakers Nvidia and Qualcomm for the future of Windows on smartphones and tablets.

You could argue that Microsoft went with ARM for its lower price point over x86, but considering the Surface RT and other first-gen Windows RT devices couldn’t even beat the price of the iPad, it’s a moot point. Plus, what good is a low price point if Windows RT devices can’t actually do anything?

Long live Wintel

In short, Microsoft could’ve easily opted to power the first Surface RT with a Clover Trail chip rather than Tegra 3. It might have been marginally more expensive, but at least the tablet wouldn’t have been a £400 paperweight that cost Microsoft a $900 million (£560 million) write-down for unsold stock. This would have set Microsoft up perfectly for a Bay Trail-powered Surface 2. Given how superbly Bay Trail performs, it’s almost unbelievable that Microsoft opted to go with Tegra 4 for the Surface 2, but we can only assume that Microsoft has signed a multi-year deal with Nvidia, or that the wizards in Redmond can see something that we can’t.

If Microsoft had stuck with x86, the company would’ve saved millions on the development of Windows RT. In terms of less tangible losses, it’s hard to put a value on the damage caused by the poor performance, lack of apps, and miscommunication about the capabilities of Windows RT devices, but it’s safe to say that it certainly hasn’t helped Microsoft’s increasingly desperate push into the mobile space.

We will probably never know why Microsoft diverged from Intel at such an important juncture. It was probably because Intel was late to the mobile party as well, losing the entire market to Samsung, Qualcomm, and other ARM chipmakers. Microsoft probably thought that Intel’s reign was over, panicked, and jumped into bed with Nvidia and Qualcomm as quickly as possible.

In classic Microsoft fashion, though, it wasn’t quick enough. If the Surface 2 and Windows 8.1 RT had come out a couple of years ago, before the iPad had completely captured the market, I think the mobile landscape would be very different today. If Microsoft had refrained from straying into the wimpy ARM undergrowth and stuck to its Intel guns, the Surface 2, with a strong x86 software ecosystem, would actually be a compelling device. Sadly, Microsoft took neither of these sensible paths, instead striking out on a weird, winding, and incredibly slow-going route with nothing more than a machete, sweaty armpits, a bouncy gait, and a maniacal grin.

Through sheer determination (read – deep pockets), Microsoft will probably get there eventually. Unfortunately, as the company and its fans know all too well, the definition of “there” is constantly being moved by its faster and nimbler competition. And thus Microsoft keeps trudging forward…

While you're here, you might also want to take a look at our examination of how Microsoft's own policies have rendered the company increasingly irrelevant.