Windows-Based ARM Laptops Are Repeating Intel’s Clover Trail Mistake

When Intel decided to push into the tablet and smartphone business back in 2011, it had a distinct vision for its x86 smartphone and tablet rollout. The goal, the company explained back then, was to establish the idea that x86 could offer premium performance and features that ARM couldn’t, then use this position to establish Intel as a preferred brand. In practice, the strategy didn’t work. The early Windows 8 tablets based on products like Clover Trail were light and had solid battery life, but they lacked the performance and storage they needed to be competitive and nobody wanted to use Windows 8 on a tablet in the first place.

They were also critically hampered by price. Clover Trail-based tablets often carried launch prices that put them in direct competition with Core i3 or even Core i5 laptops. There were a number of reasons why Intel’s approach didn’t work, but the bottom line is this: Weak tablets and convertibles, high prices, and an uncertain software situation in which consumers aren’t sure if they can use their applications on your hardware platform are not a recipe for strong product sales. I thought the industry had learned that lesson after the way Intel’s Windows tablet efforts flamed out back in 2011 – 2013, which is why it’s so surprising to see ARM vendors merrily copying some of Intel’s worst habits.

THG has the scoop on this, detailing how the recent Samsung Galaxy Book 2 they’ve reviewed struggles far too much in Windows to justify its $1,000 price tag. Avram Piltch writes:

During my tests, its Snapdragon 850 chip struggled with Chrome, the world’s most popular web browser while its paltry 4GB of RAM made multitasking painful. When I surfed the web with over a dozen tabs open and a YouTube video streaming in another window, there was significant lag. Every time I changed tabs, I had to wait a couple of seconds for the page to redraw, even though the pages were finished loading. At times, the system locked up for a few seconds.

To see how a similarly-priced, Intel-powered computer would handle the same workload, I tried opening over a dozen tabs and playing video on a Surface Pro 6 with a Core i5 CPU and 8GB of RAM. Not only was there no perceptible lag when changing tabs, but web pages loaded a lot faster, even though both the Surface and the Galaxy Book 2 were on the same Wi-Fi network.

Excel test run natively, ARM versus x86. Image and data by Tom’s Hardware

This isn’t so different from our tests on the Samsung Ativ six years ago, when we ran into similar problems on Clover Trail. Despite featuring a Qualcomm Snapdragon 850 instead of the more anemic Snapdragon 835, Samsung’s Galaxy Book 2 performance isn’t good, even in ARM apps that aren’t running in emulated Win32 mode. When actually using emulation, performance is even worse. The author winds up questioning “Who is this for?” which neatly echoes the question I raised with the Samsung Ativ six years ago, namely, “Who’s supposed to want this?”

Are These Fixable Problems?

I’m not claiming that the Ativ of 2012 and the Galaxy Book 2 are in the same shape. Reading THG’s review, it’s clear that the Galaxy Book 2 is a much better system overall than any of the early Windows 8 hardware we saw back in 2012. But the fact that so many of the same types of problems are popping up as far as performance and basic compatibility means ARM on Windows in 2018 looks a lot more like a mixture of Surface RT and Clover Trail issues in 2012 than anyone at Microsoft or ARM should be comfortable with. The Snapdragon 850 is faster than the Snapdragon 835, but the 15 percent gain THG reports and the general sluggishness of the system suggest those gains aren’t nearly enough to build a genuinely desirable experience. Desktop gaming isn’t happening and not all Win32 apps are compatible with the emulation MS provides (OpenVPN isn’t, because its TAP driver isn’t supported on ARM).

These systems might move at $500 – $600, though I’m not actually sure of that — I’m not sure anyone needs to pay hard cash for a compromised computing experience, and that seems to be what ARM and Microsoft are capable of offering right now relative to what you’ll get with a standard x86 machine. The prospect of an $800 – $1000 computer with many of the same flaws and compromises of six years ago and fundamentally unattractive pricing relative to what you can buy elsewhere isn’t much different than it was then. But it’s genuinely surprising to see Samsung walking into a strategy that failed so completely for one of its largest competitors. In 2012, a mixture of software, hardware, and technology issues meant that Intel’s then high-end tablet platform wasn’t fast enough to justify its price premium, the software wasn’t ready for Windows RT, and the entire cost structure of the product stack for both ultra-mobile experiences was broken, which is part of why both Windows RT and Intel’s larger tablet push in Windows with x86 failed.

If companies are serious about using ARM hardware to try and break the x86/Windows monopoly, they’re going to need to set prices that make sense for the actual level of provided performance and overall capability, then raise those prices over time as the level of capability improves. It’s either that or watch Windows on ARM 2.0 play out exactly the same way, two times in a row.

Now Read: New Details Leak on PC-Focused Snapdragon 1000, Microsoft’s Windows on ARM Effort is Seriously Limited, and Qualcomm Unveils Snapdragon 850, Explicitly Aimed at New PCs