At Computex, Qualcomm announced it would build Windows 10 systems with major traditional PC OEMs. That’s a major step forward for the company, which flirted with building mobile PCs during the Windows 8 era when Windows RT (née Windows on ARM) was in development. The bulk of that market, such as it was, appears to have mostly gone to Nvidia and its Tegra 3 and Tegra 4 SoCs. Qualcomm has never worked with any of the major PC OEMs to build Windows systems, and the Snapdragon 835’s gigabit wireless support could help it stand out compared with other Windows devices in the same approximate market.
Qualcomm is claiming that these devices can deliver up to 50 percent more battery life than equivalent x86 devices, with 4-5x longer stand-by times. They’ll also build on the Connected Standby feature found in Windows PCs already, with the ability to wake up, check email or update applications, and then return to sleep.
The devices will contain Qualcomm’s X16 LTE modem, which can offer up to gigabit performance. Having an LTE modem inside a Windows portable also means you won’t have to worry about Wi-Fi to get online. That said, the mobile data plans telcos like ATT and Verizon offer tend to have eye-watering costs per gigabyte if you go over your plan allotment, and it’s arguably easier to do that on PC than on a mobile device, depending on how you use it.
At Computex, Qualcomm showed off its own circuit board designs compared with those it claimed would be required for an x86-equivalent design. The Qualcomm board is significantly smaller (50.2cm versus 98.1cm).
To Boldly Go Where We’ve Been Trying to Go For Years
Let’s get a few things out of the way up front. It’s entirely possible that close collaboration between Qualcomm and Microsoft has been to the mutual benefit of both companies. Qualcomm knows how to build great radios, Microsoft knows Windows 10 inside and out, and it’s not hard to imagine the two companies working together to create a great hardware platform. There are two major features Qualcomm will need to offer, however, to make this platform attractive to existing x86 customers.
First, emulation performance has to be strong. Back when Windows RT debuted, plenty of users wanted Windows x86 compatibility on the platform, but the old Cortex-A9 simply wasn’t strong enough to provide a satisfactory experience. The Snapdragon 835 is vastly more powerful and Microsoft has had years to work on the problem, so getting emulation working well shouldn’t be impossible. It will, however, have an impact on both battery life and overall performance compared to a native x86 chip.
It’s possible that one reason Qualcomm still feels confident about winning these comparisons because ARM has continued to iterate on its CPU designs, while Intel’s last major refresh of its smartphone and tablet division was back in 2012, with the launch of Bay Trail. While 14nm Cherry Trail chips launched as recently as Q1 2016, these were die-shrunk Bay Trail parts, not a brand-new architecture.
Second, Qualcomm needs to be able to deliver on its battery life claims. Years ago, this would have been a given. New processes (10nm in this case) meant higher clocks, lower power consumption, and improved battery life, all at the same time. Today, these gains tend to be much more workload-dependent. Doing a lot of video encoding and decoding? Then new enhancements in that area from Intel, AMD, or ARM will give you much better battery life and/or improved performance. Still running garden-variety applications? Not so much. And that’s before we talk about how browser choice impacts battery life.
I’ll be surprised if Qualcomm can deliver massive across-the-board improvements on the above front. That’s not because I doubt the quality of the company’s engineering, but because studies of ISA efficiency have generally found that above the microcontroller and embedded markets, it doesn’t really matter which ISA you choose. ARM and x86 have been found to compete fairly well in that regard.
Finally, I’m uncertain how much these new announcements will fundamentally reshape the PC market. Back in 2011, the IT industry (including this author) expected a major brawl between Intel and ARM as the former shrank down and headed for smartphones while the latter grew up and headed for laptops. Windows on ARM / Windows RT was actually seen as a product that was critical to Microsoft’s long-term future, if you can believe it. Despite oceans of virtual ink spilled on the topic, in the long run, not much happened. Windows RT failed, Intel remained focused on PCs (despite spending ten billion dollars or more trying to break into mobile), and ARM stayed supreme in the tablet and smartphone space.
Clearly, Microsoft wants to take another stab at the Windows on ARM equation, but if the past is any precedent, it’s facing a steep uphill climb. Best case, ARM opens a new front against Intel in mobile while AMD goes after it in data centers and mainstream computing, and Microsoft sees its own market share rise in the total OS market. Worst case, Intel fends off both challenges, the Windows on ARM initiative collapses (again), and we all keep using Android and iOS devices with one type of CPU, and desktop and laptop PCs on another.