Microsoft’s Windows continues to dominate the desktop operating system space, but in recent years Chrome OS has taken away a significant chunk of market share, particularly in entry-level laptops and computers designed for students and classrooms. Meanwhile Android and iOS dominate the smartphone and tablet space.
Over the past few years Microsoft has made several attempts to offer a stripped-down version of Windows that would be more competitive on entry-level hardware with limited success. Windows RT is dead. Windows 10 S is… basically a crippled version of Windows.
So what’s next? Windows Lite, apparently.
Rumors have been making the rounds for a while that Microsoft was building a new operating system that would be capable of running on entry-level hardware while offering a simpler user experience.
Now Petri’s Brad Sams reports that the so-called “Lite OS” is under active development, and that Microsoft plans to expand its testing by this summer. The company could officially introduce Windows Lite during its Build developer conference in May.
Sams also created a mockup image that gives us an idea of what Windows 10 Lite looks like. There’s a simplified taskbar with app icons in the center and a clock on the right. And there’s an app launcher that looks a bit like the ones you’d find in Android or Chrome OS, with a search bar at the top, suggested apps below it, and a section for pinned apps below that. There’s also a Documents tab, suggesting you’d be able to browse and search for apps and docs from the same launcher.
According to Sams, the operating system does include some legacy features such as File Explorer and support for running apps in windows that can be resized and moved. But Microsoft is reportedly working to make Windows Lite easier to use and to maintain than other versions of Windows.
Right now there’s apparently no built-in support for running Win32 desktop applications. You can only use Universal Windows Platform apps (like those available from the Microsoft Store) or Progressive Web apps. But Microsoft is investigating the possibility of adding support for Win32 apps, possibly by sticking them into containers and limiting their interaction with the operating system (similar to the way Chrome OS handles third-party apps), which would theoretically tighten security and prevent installed apps from slowing down a computer’s boot process or other functions.
Microsoft hasn’t confirmed any of this yet, and even if everything in the report is accurate there’s a chance that Microsoft could change its plans before Windows Lite is released (or announced).
It’s also not entirely clear to me if people want a version of Windows that acts less like Windows and more like Chrome OS. You can already run Windows 10 on some pretty inexpensive hardware… it’d just be nice if Microsoft insisted that OEMs use at least 64GB or 128GB of storage on entry-level Windows devices in order to reduce the risk of Windows 10 updates failing due to limited disk space.
Update: The Verge has its own sources that confirm Windows Lite is coming, and that the mockup from Brad Sams looks pretty accurate. One new bit of information from The Verge’s reporting is that Microsoft may initially be targeting dual-screen devices, before moving to support cheap Chromebook-like computers.
Windows is already dead to 96.3% of the market place and, as of 2017, should be a faded memory ‘just like Intel. The ARM Architecture looses 14 to 18% of its speed via the SPi Eathernet and G5 Wifi routines which Microsoft is known for. Currently ARM Programs are considered Bloat-Ware if they are over 18 gigabytes for the entire Software Bundle; Windows Lite is forcasted to run at 96 Gigabytes plus will be chewing-up Processor Speeds at up to 18% More via The Internet/Cloud Usages.
Many verities of DOS and Win XP & 7 are already in use world wide on the ARM Platforms and people are quite happy with this. Windows gave up their support of these systems years ago on the Intel x86 Platform and quite frankly — we really do not need a Programmers Flop like Windows 10 encrouching upon happy campers world wide.
Microsoft had its chance to separate itself from The Gamers Domain ‘years ago and didn’t get the hint when AMD no longer supported non-GPU Processing. Therefore, it was once time for MicroSoft to either Stay with the Gamers Domain or die a long and expensive Death trying stay with Business and Management Solutions after the 2004 Processor Wars Battle. PipeLine Gaming is a joke so, Windows and Gaming shall both follow Intels lead in a very agonizing Grave however, unlike Intel; Microsoft will not be missed.
Funny how 32gb Windows laptops always run out of storage, but 16gb Chromebooks don’t. One problem is the Windows recovery partition taking up almost half the available storage. Windows XP could easily run with 32gb of storage. Microsoft should have stuck with XP!
Thanks for mentioning about the absurdity of devices with too small of storage to deal with Windows 10 updates. It’s shameful. I’ve mentioned this before, but…the way Microsoft stops doing this is a law suit. They can’t plead not their problem. Afterall, Windows 10 comes on the devices, pre-installed. Fact is you can’t really have Windows 10 on a 32GB device but sure enough those are sold and on the market. Law suit. That’s the solution. They will understand that. Amazing that Microsoft is still trying to create half-assed versions of its OS. How many failures?
Microsoft imposes those rules on OEMs in order for them to get Windows Licenses for $15 instead of $50 and better compete with ChromeOS devices.
– Displays no larger than 14.1-inches,
– Low-end CPU (Bay Trail, Cherry Trail, Braswell, or Apollo Lake),
– no more than 4GB RAM,
– mandatory use of an SSD/eMMC with no more than 32GB of storage space, and no option of a hard drive or optical drive
These things are absolutely dreadful.
Windows Lite is going to be DOA… because Microsoft is going to do the same half-hearted attempt at producing a lightweight version of Windows. At the very least they should gut out the Enterprise-level subsystems that aren’t needed for the average consumer. But that will require significant regression testing to determine what works and what doesn’t. They don’t want to invest any serious money in that so they’ll tweak a few registry entries and call it good.
You’re right. The management’s position is this:
We’ve dominated the market, so we just want to retire and rake the money in. When people ask, we will throw money at marketing to try to convince our investors that we have grand plans and we will be shaping the future of technology and other bullshut.
Probably the bare minimum OS to run Office 360 and Internet explorer. I would expect them to give it away for free (just like ChromeOS and Android).
My biggest problem with RT and all the crippled versions of Windows was not that they sucked. IT’s that they didn’t stick around long enough to not suck. Every OS sucks at the beginning. Android looked like a horrible mess up till 4.1 (and beyond), iOS didn’t even have a wallpaper or multitasking until the iPhone 4. Chrome OS is just beginning to resemble an OS and not a Kiosk. It took 4-5 years for each of them to become good, or at least “not as bad”. Meanwhile Windows RT existed for about 2 years and about 10 devices tops. At some point in the future we will have to shred legacy x86 support, that’s not up for debate, it’s been a major problem for almost a decade now that cripples new chips. We should design the PC from the ground up to be more efficient, because we can no longer just throw more cores or higher clockspeed to increase performance anymore. New CPU architecture (ARM?), new storage scheme (memristor/Optane?) and a new OS that leaves the baggage behind. Yes, we will need devices for legacy support for industries and professionals. We can just include legacy CPU cores for those cases. A personal PC that you use for browsing and working in an office environment can be powered by the new tech. Want to run a 20 years old accounting software? Well, here is a CPU with 16 ARM cores and 2 legacy x86 cores dedicated for legacy emulation. But they have to stick to the plan, not for 2 years, not for 4-5, this transition will take a decade. The question is not if it will happen, it’s that who will be the one who makes this happen? Microsoft has the resources to go through with it, but apparently not the patience. Apple is rumored to do just this in a few years. They were never concerned about backwards compatibility, so they can do it easily, but they no longer lead the trends in the PC market. Google gets this, but chromebooks are not a slice big enough to merit anything just yet. And if nobody is willing to do it, then it’s time for linux to step up.
this will never happen, Intel and Arm will not mix… Instead the future of laptops is Windows on Arm, with emulation for legacy software. Not an immediate futures, but more a 5-10 year one. Same for Macs, Apple is moving towards its own Arm chips. Intel CPUs will only remain dominant in desktops eventually.
You know, back in the Amiga days we put 80486 cards in our Amiga to get Dos and Windows running native (it was called a Golden Gate card I think). There were cards for 68k macs of the day for the same reason. In fact you could run Win95 from within MacOS with all the hardware acceleration. This was the early ’90-ies. I’m pretty sure that if Intel wanted to, they could easily stick an x86 core next to some (licenced) ARM cores.
Just plug an intel stick in the Thunderbolt 3 port if and when you need Wintel.
Intel has an ARM dev license, they could join the party anytime if they wanted to.
“At some point in the future we will have to shred legacy x86 support, that’s not up for debate, it’s been a major problem for almost a decade now that cripples new chips.”
“And if nobody is willing to do it, then it’s time for linux to step up.”
As far as I know, as much the Linux desktop exists, it exists on x86. Where do you see the future here?
Linux is usually the first thing they port to any new platform. It’s extremely scalable and it’s very fast to adopt it to a new system. It can power a couple MHz single core CPU of a supercomputer with thousands of CPUs. Heck, one time a guy got it to boot on a microcontroller with some extra RAM attached.
But that’s besides the point. What I think the future is is a new kind of PC architecture, derived from the ARM style of CPUs. Why do I think it’s timely? Because we hit a wall with current development. Currently there is nowhere to go forward with regular x86. We can’t increase the clockspeed significantly, there never has been any systems running well beyond 4GHz. We can’t increase the core count, because not every algorithm can benefit from parallel execution and we can’t really shrink the transistors anymore, Intel struggles to reach 7nm and it seems the absolute limit will be 4nm.
So where can you go from here? If you can’t make the clock run faster, can’t increase the core count and can’t fab on a better tech, then the only way forward is to optimize what you currently have. And one way of doing so is to get rid of everything that is not or just rarely used. And that is legacy x86.
The current x86 CPUs are designed in a way that they can even execute 8 bit code. In fact, the basic code, the one the CPU expects is the 16 bit 8086 code. So anything beyond that takes some extra from the CPU. Think of a simple multiplication, you load your values to AX and BX and get your results in the DX:AX pair. Now you have to write code to handle the fact your result is in two registers. Or, you can use an extended MUL function and load and get values in 64 bit registers, but the tradeoff is a longer execution. In a new architecture you can just make a single MUL function that expects 64 bit and gives 64 bit. Or 128 bit, whatever.
Secondly, we now have memristors. One of the big problems with PCs that had to be solved from the very early ages of IT is that RAM is volatile. There always had to be some weird refresh routines to keep the information in RAM and varying levels of different kind of RAM trying to cache it as effectively as possible. Currently there are about 7 levels of RAM in a single PC -> the storage medium, the cache of the storage medium, RAM itself, L3,2,1 caches and the registers in the CPU. Every single executed bit has to move through all this. With the memristor the RAM and the storage medium can be the same, and if they can get fast enough, they can probably drop the L3/2 cache as well. Or maybe they can be on the same silicon with the CPU cores and the new architecture could just operate directly on the memristor! Just think about it, there would be no more RAM limit, you could use whatever space you have on the memristor for either storage or execution, dynamically scaling between the two.
Third, we will need legacy support to some extent. No business would be too happy to rewrite their expensive business software for a new architecture. Heck, even smaller changes are hard to swallow for some business, just the other day I’ve seen an ATM running on OS/2, and OS/2 died more than 20 years ago. And, for this I propose to integrate a legacy x86 core into the new architecture, only for executing these legacy softwares on the new architecture. Yes, purely software emulation is possible, but maybe for the first few years, until the transition is complete they could mix in something small.
Fourth, the software. Did you know, you can run 16 bit windows apps on Windows 10? The API is still there. You never use it, I never use it, maybe a sawing mill somewhere has an industrial machine that uses it, but it doesn’t have to be part of the OS anymore for 99.9%. This new windows lite don’t need to have this. And if you take out all the legacy layers from windows – all the .NET packages, and directX 9.0 support from 2002 and such you could get an OS that is just the fraction of the size of the current one, and can run almost all apps the majority of people use, and maybe keep around a “legacy edition” for those businesses that need the old code to work.
TL;DR: a new, lighter windows would be good, but it needs to happen along with a new architecture and Microsoft needs to stick around with the plan for a long time. It can’t happen in a single step, which is what killed RT in my opinion, and what is going against Windows on ARM however good that sounds.
Thanks, that was cool!
I don’t see why. The venerable PC platform has been taking less and less of the overall computing pie for years now, and with mobile devices and cloud computing competing for the dollars of the consumer, it will continue to (slowly) be less relevant in the years ahead. Killing off x86 support is essentially the same as killing the PC, and that’s already happening.
And let’s face it, even entry level processors from AMD and Intel have more than enough processing power for all but the most demanding of PC users these days. Other than PC gamers and content creators, nobody is being bottlenecked by today’s CPU architecture.
In mobile and cloud computing, the x86 architecture is or has already been supplanted in many areas. AI learning tech mostly uses specialized processors already, and of course the entire mobile market has been dominated by ARM almost since the beginning.
I would suggest that x86 and legacy apps have still not reached their full potential, so it’s not worth dropping it all together. x86 is power hungry and that’s a limiting factor, but there are lithium sulfur batteries already available with admittedly shorter life spans but higher energy density.
The real reason why companies are afraid to drop x86 is most places handles legacy apps like s*hit while relying on them heavily. I’ve been in a few projects, and usually nobody is there who was part of the original team, nobody understands how the apps work, the documentation is lost of so out of date that it’s unuseable, the codebase is an unmaintained mess with patches and hacks and ‘clever solutions’ that literally only work on one very specific and old version of some runtime environment. I’ve seen this in the private sector and in governemnt sector as well. Sometimes companies stock up old machines just to ensure their old apps keep working. It’s horrible and sooner or later they’ll all have to figure out what to do, and the longer they wait the worst it gets. There was this mall that had their car park barriers operated by a Pentium II machine with Win 98SE and some cobbled together adapter on the serial port. The whole thing could be run from a RasPi (or maybe even an Arduino), yet they need to keep alive a decades old PC that fails at least once a month, because they don’t want to pay for replacing it, since “it works”.
On the other hand I’ve seen hospitals using old machines simply because the medical devices connect to those specific softwares that need legacy support (Win XP for example). A perfectly fine X-Ray machine, only its drivers were written for NT4.0, and cases like this. This I can understand, and for such users a legacy core integrated into the new architecture would be a solution. A new, fast, secure system, the new OS and a virtualized environment with hardware acceleration for these legacy usecases.
True that companies who deploy equipment relying on legacy apps and equipment can be an administrative nightmare. However, let’s talk opportunity and open standards. Over the years we’ve seen the price for entry level full-Windows PCs drop from +$1K to ~$100. I’d say some people and applications are coming to the table that haven’t been there before. Granted, this same thing is happening on the ARM side…
This is true. I watched Chrome OS for 4 years or more before I finally bought a Chrome OS device. MS never seems sticks with anything these days. Windows phone had several complete re-writes of how their apps worked. breaking all new apps on all old phones if they got the upgrade, several times, in the name of progress. never worked out
Comments are closed.