Skip to main content

Why Microsoft might go with an Intel chip in a new budget 10-inch Surface

Microsoft Surface logo
Microsoft Surface logo (Image credit: Daniel Rubino / Windows Central)

Earlier this week, a new report emerged online that claimed Microsoft is planning to ship a new budget 10-inch Surface device later this year. The last time Microsoft updated its "budget" line of Surface tablets was in 2014 with the Surface 3, which shipped with Windows 8.1 and was powered by an Intel ATOM processor. A lot has changed since then; ATOM is more or less no more, and Microsoft has started investing in bringing Windows 10 to ARM processors.

Microsoft's budget line of Surface tablets have never been power-houses, and that's because they're not supposed to be. The Surface RT, Surface 2 and Surface 3 were all aimed at doing light productivity work and normal consumer-related activities. If you needed a device for video editing or a more heavy workload, you'd spend a little more and get a Surface Pro. But the standard, non-pro line of Surface tablets was always great for low-powered chips.

So, a natural successor to the Surface 3 would be a Surface 4 powered by ARM, right? Well, apparently not. According to the report, the next budget Surface will be powered by Intel, and Mary-Jo Foley hears this will be an Intel M chip. An Intel M-based chip is way better than an ATOM one, that's for sure, but with Microsoft betting big on ARM, why would the company opt for Intel in its own budget, low-powered line of Surface tablets over Snapdragon? Surely, this is where an ARM processor makes the most sense in Microsoft's Surface line-up, right?

I personally think that yes, an ARM processor would make more sense here. However, there's likely good reason behind Microsoft's choice of going with an Intel chip instead. It's possible that this new budget Surface will be marketed primarily at students and education, just like the Surface 3 was. In most cases, students don't want or care about LTE in their PCs, and unfortunately, Qualcomm requires all Windows devices that ship with a Snapdragon processor to have LTE capabilities. And Microsoft likely wants to sell non-LTE variants of this device.

LTE or no LTE

As such, it's important for Microsoft to be able to sell a non-LTE version of its budget Surface tablets because that's the version that most people will want because of its lower price. The report does claim that an LTE model will be available but doesn't specify whether that version will be ARM powered. Regardless, assuming there won't be an ARM version, the LTE version of this new budget Surface can still be classed under the "Always Connected PCs" (ACPC) moniker Microsoft is pushing. Even without ARM, this new budget Surface will still be a modern PC.

The other probable reason for going with Intel over an ARM processor is likely due to the infancy that is Windows 10's x86 emulation. Right now, being able to emulate x86 on ARM is more of a safety net, put in there for emergencies when you really need to use it. You shouldn't be buying an ARM device if you're planning to use only emulated apps, because that experience will be slow and bad.

Microsoft's Surface 3 was a great little tablet when it was first released. In fact, for a while it was one of the more popular Surface devices on the market. The only issue with it today is that it aged rather poorly. The Intel ATOM chip it uses is a dog, and things like Micro-USB over Microsoft's own charging port or even USB-C is a no-go in 2018. All a new budget Surface needs to do update the processor, add USB-C, maybe shrink the bezels a little bit, and update the accessories with new designs that were introduced with the Surface Pro 4.

A $400 Surface is nothing to laugh about. That price alone may be enough to make this new budget Surface the most popular Surface to date. Let's hope Microsoft can nail the execution.

Zac Bowden
Senior Editor

Zac Bowden is a Senior Editor at Windows Central. Bringing you exclusive coverage into the world of Windows 10 on PCs, tablets, phones, and more. Also an avid collector of rare Microsoft prototype devices! Keep in touch on Twitter: @zacbowden.

  • Here here to an updated Surface, whatever is under the hood. I didn't know that Qualcomm required SD tablets to have a LTE modem as well. It helps explain why the HP X2 is relatively pricey.
  • No, LTE does NOT explain the pricing of Qualcomm SD tablets. The modem is already part of the SOC, which only costs somewhere about 50 dollars a piece and has EVERYTHING on it, like GPU, networking, etc.... 😊
  • > I didn't know that Qualcomm required SD tablets to have a LTE modem as well. They certainly didn't use to: neither Samsung Galaxy Tab S2 (SD652), nor Samsung Galaxy Tab S3 (SD820) have one. I wonder where the information WRT this requirement came from.
  • The Snapdragon chips work on a licensing model, as I understand it. Everything is built into the chip, but if you want to activate the cellular radio, you have to pay more. There are also some additional costs related to including an antenna that will work with the dozens of bands that a modern cellular radio may communicate on, as well as FCC testing. It was likely a deal struck by Microsoft and Qualcomm to increase the average selling price by guaranteeing that the cellular radio be activated on every device sold. Unfortunately, they priced themselves right out of the market.
  • Ooh that would be nice to try out.
  • I need one now, my surface 3 is just 3 year today. :-)
  • It's only about performance. ARM is not there yet.
  • Tell that to Apple... they apparently didn't get the memo. They're producing low powered ARM chips that rival i5 based desktop Intel chips.
  • well its Windows on ARM that is not there yet.
  • No. Windows on ARM is fine. It's x86 apps (and drivers) that have not yet been ported to ARM that lessen the experience.
  • It's really the whole ecosystem. Going ARM really means that you'd have to run most of the apps you want under emulation right now. Developers have little incentive to code for both Intel and ARM on Windows right now. As long as Intel can remain reasonably competitive, this shouldn't be a big problem for Microsoft. It's clear that Microsoft wants to go ARM, it's just that they can't really pull it off with their development community. Microsoft needs a break out success with an ARM based product that will give developers incentive to move.
  • With UWP devs can target ARM/x86/x64, existing apps would be the issue.
  • Developers do not have to code for ARM at all, they code against an API which is available for both x86 and ARM. You just compile the very same code for ARM and x86 targets.
    On the second part you are right, Microsoft needs a low cost ARM tablet - so the Surface 4 would be the perfect opportunity.
  • They need Chinese OEM's to jump in with MediaTek chips, or Huawei, Xiaomi and Samsung with their own ARM chips. When there are enough of these devices on the market, developers will have no choice but to compile for ARM64.
  • Why should they if not even Microsoft releases an ARM tablet? The original idea of Surface was to show OEMs how to do it right, this spirit has long been lost. Microsoft is shooting themselfs in the foot with such short-term thinking.
  • No, it's just Chrome. Don't install that spyware and the devices run just fine.
  • Edge is just as much spy ware as chrome is covfefe.
  • Why not use Firefox or other Forks of Firefox instead?
  • They aren't running legacy Win32 software on them either. In education you will have to deal with legacy support.
  • Look at Build 2018 session named "BRK2438 Windows 10 on ARM for developers", on existing hardware (Snapdragon 835), they showed a few things:
    - VLC (recompiled) playing 4K video smoothly
    - OpenVPN (recompiled) with a re-compiled ARM64 driver, zero line of code change
    - Asphalt 8 (x86 emulated, untouched from Windows Store), play smoothly, around 35:40 in the video Of course, performance is bad on Photoshop and Chrome. But I think it's relatively easy to recompile Chrome to work great on ARM64.
  • Ahem... both VLC and OpenVPN (complete with underlying libraries) were *ported* to ARM quite a while ago. What was "recompiled with zero line of code change" was just a Windows-specific lipstick, someone put on the huge amount of underlying code. Nice sleight of hand on Microsoft's part, but hardly proving much of anything.
  • There is no porting involved. You just recompile. What is so hard to understand about the fact that the ARM devices are API-complete?
  • A lot of things are hard to understand, apparently. I would suggest you plug word "endianness" into your search engine of choice and start from there. If you are unimpressed, check out VLC code *and all the underlying libraries* and 'grep \#ifdef' throughout the tree. Oh, yeah, search for '.S' files while you are at it -- from my (vague) recollection, at least libjpeg had Intel assembly language bits here and there...
  • Lol, apparently you have no idea what you are talking about. How many apps did you re-compile to ARM64? So no personal experience? I guessed that... Regarding your particular points: Endianness: both ARM and Intel are little endian machines - otherwise you had bigger issues when re-compiling. The *.s files are conditionally compiled and are typically target specific. So if the target does not match, they are just not considered for building. Besides i did compile both libjpeg and ffmpeg for Windows ARM64 without any issues - so there goes your argument.
  • > How many apps did you re-compile to ARM64? You are absolutely right -- I do not get to *re-compile* all that much -- I wish I was that lucky. OTOH Microsoft was able to *recompile* the stuff because the work of porting (either creating target-specific variants and bracketing them with conditional compilation or providing, potentially suboptimal, target-neutral variants) was done by someone else before. Makes for the cool demo. Back to JPEG library: $ find libjpeg-turbo-1.5.3 -type f -name "*.S"
    libjpeg-turbo-1.5.3/simd/jsimd_mips_dspr2.S You can open jsimd_arm64_neon.S file and in the few lines from the top you will find people who worked hard so Microsoft can *just recompile*. Admittedly, there is a lot of open-source code that has been ported to ARM for the variety of target OSes, so things are not that bad. Proprietary applications, OTOH, especially the ones that were strictly Windows-based, will be much more challenging. Do not get me wrong -- I wish this device every bit of success, if for no other reasons that people, who write and support cross-platform code for a living, could be gainfully employed :)
  • Well yeah that is the turbo library, i used the normal variant without assembly from (, but even the turbo library has the required NEON code, as you correctly pointed out.
    But that´s not even my point, most projects, which have an assembly code path also offer an equivalent C-code path and thus compile just fine - without any work by anyone. In fact i have seen projects, which just offer an SSE4.2 or AVX2 code path and fall back to C even on Intel CPUs without these capabilities. I have yet to see code without fallback option. As i did mention in another post, there are a few hick-ups here an there. As example in MAME source you find the following lines: // Detect with _LITTLE_ENDIAN and _BIG_ENDIAN macro // Detect with architecture macros
    # elif defined(__sparc) || defined(__sparc__) || defined(_POWER) || defined(__powerpc__) || defined(__ppc__) || defined(__hpux) || defined(__hppa) || defined(_MIPSEB) || defined(_POWER) || defined(__s390__)
    # elif defined(__i386__) || defined(__alpha__) || defined(__ia64) || defined(__ia64__) || defined(_M_IX86) || defined(_M_IA64) || defined(_M_ALPHA) || defined(__amd64) || defined(__amd64__) || defined(_M_AMD64) || defined(__x86_64) || defined(__x86_64__) || defined(_M_X64) || defined(__bfin__)
    # elif defined(_MSC_VER) && defined(_M_ARM)
    # else
    # error Unknown machine endianess detected. User needs to define RAPIDJSON_ENDIAN.
    # endif
    #endif // RAPIDJSON_ENDIAN Problem was, that MSVC defines _M_ARM for 32 bit ARM and _M_ARM64 for 64 bit ARM. Since i was compiling for 64 bit i had to modify the macro accordingly - which was trivial - and gives you a picture, what typically small problems you potentially run into. But in the big picture, everything was compiling well without me being an expert on MAME code.
    Now if i would have to port the code to say iPhone, this would be a real challenge (and would take months instead of a day)
    Therefore a make a big difference between compiling for same platform but different target and really porting to a different platform. Only the latter one i would really call porting.
  • > # elif defined(_MSC_VER) && defined(_M_ARM) So the work of, at the very least, ensuring that this code will run on the ARM was done by someone *before* you came along and modified the macro. The former I call "porting", the latter I do not. My point is that a lot of open source stuff is already ported to variety of CPU, including ARM (thank you, Android) and, likely, did not originate on Intel to start with. OTOH the things, created for Windows in the last 20+ years, were exclusively targeting Intel (yes, I know about DEC Aplha and no, I have not seen it outside of US Navy since 1998). I guess, I just wished Microsoft gave appropriate credit to people, who did the heavy lifting and did not deceive budding developers with the promises of "recompile"... 'nuff said.
  • The one before me just added the _M_ARM macro and i did add the _M_ARM64 macro - both is not porting as we did not touched a single line of actual code. We just did set the appropriate environment for compilation - as i said a small hick-up when after this change a few hundred thousand lines of code just compiled - it not even worth mentioning to be honest.
    Why are you assuming that there is heavy lifting involved at all? Its just C/C++-code compiled against the Win32 API - there is no heavy lifting at all necessary. Its completely irrelevant if the target is x64 or ARM64 as long as the API is there. The above example is also a very specific and rare case, because the original developer tried to program endianness neutral and just missed _M_ARM64 in single macro. (which is understandable since _M_ARM64 is MSVC specific). The vast majority of existing code do not even distinguish between different targets and is totally target neutral.
    So Microsoft is totally right, in most cases you just press the compile button and you are done (assuming the project is already Win32 or UWP of course)
    Besides, all the specific changes done for Android on _all_ projects i was working on are related to Android API and never related to ARM - not a single line of code.
    I can only repeat myself - there is no porting involved.
  • > Why are you assuming that there is heavy lifting involved at all? From personal experience, I guess. But, I will take your word that ARM (as implemented by PC vendors) is sufficiently similar to Intel in endianness, sign expansion, unaligned access, floating point handling and million other little things. Thank you. > Besides, all the specific changes done for Android on _all_ projects i was working on are related to Android API and never related to ARM - not a single line of code. Were you really calling Android API from the native code? That is as cool as it is masochistic...
  • The situation isn't that simple. As a developer of any complex software you will use libraries from third parties rather than develop every component of a large app yourself. This means you need them all compiled to ARM along with your app and may not have the source code. Also a certain way of coding a method may be fast in x86 and slow in ARM or vice versa so performance testing and optimization is still required. WoA is doing a great job using binary translation but its not perfect there is always an overhead and a straight recompile to ARM is usually more complex than it sounds.
  • Again, this most be come from someone, who has not tried to compile a reasonably complex project to ARM64. Yes, you need to compiled the libraries you depend on as well. But that is it. And yes, you might need to fix a few things here and there - mostly related to the fact that the original programmer did wrong assumptions.
    Example: I did compile MAME for Win32 64bit ARM (a project which has almost 200 subprojects and quite a few external dependencies) In a few places there was the assumption that if the platform was Win32 and the Machine was 64 bit, then it must be an Intel machine. This assumption was wrong and i just fixed it.
    Other example: WinFellow - an Amiga emulator. This was a bigger challenge - since WoA does not support DirectInput - so i had to rewrite the Joystick and Keyboard driver to use XInput instead - ok took be 2 days because compilation itself was a non-issue.
  • The performance is there, it's the software that isn't up to snuff.
  • EMEA regions, probably like EUR 1000,- Lets hope that won't be the case.
  • Anyone still sticking around in the Islamic Republic of Europe who is not from a third world country is out of their damned mind.
  • And why not AMD?
  • AMD has nothing to offer like a Core M CPU for a small tablet. A Ryzen U wouldn't work.
  • Your statement is false Apfel
  • Because only GOD knows when AMD will make quality fanless chip with low tdp. During intel baytrail time AMD demoed fanless "mullins" chip but nobody used it since then they are quite about it.
  • AMD has plenty to offer, see my comment below
  • Isn't the 835 a complete SOC with built in LTE modem? As in one of the benefits thats been touted as the advantage of WoA? Even if it isn't, it seems it would be cheaper to eat the cost of the modem vs the cost of a Core M chip. And why wouldn't students want LTE on their devices?
  • And all day battery life and instant on
  • Battery life would definitely be a plus for students but LTE less so since they'll have lots of wifi access
  • Students are unlikely to want to pay a monthly bill for cellular connectivity when every place they go to do work is already covered by free Wi-Fi. I'm not so sure that the actual cost of using a Snapdragon 845 is much less than a Core M. I've seen 2-in-1 laptops with Core M priced below the $700-1000 range that SD 845 phones occupy. Also, it's not outside the realm of possibility that Intel would give Microsoft a discount to stop them from going all-in with Qualcomm. As Qualcomm chips continue to close the performance gap (they haven't yet), Intel will have to drop prices on Core M. Maybe this is the first step.
  • Intel only dropping prices as long as they have to. Same happened with Surface 3, when Intel ran the Contra-Revenue program for Atom SoCs - and Microsoft swallowed the bait. Intel will do anything to stop penetration of ARM SoC into the Windows ecosystem - they do know they have the inferior architecture.
  • Yeah, battery life will suck with Core M so it won't be useful to replace my atom based tablet with.. Waiting 6 years to find a replacement :-(
  • First and formost windows 10 needs a better tablet mode and tablet expreience for the next surface device. Timeline and sets are not enough. There are enough suggestions and pointers of improvement that microsoft can easily adress to improve this. Many of them relate to half finished features.
  • IMO its good now. Only issue earlier was the large keyboard but now since mobile keyboard is also an option, typing became easier.
  • Is that a chrome sticker in the thumbnail?
  • How would this fall under the "ACPC" moniker if the tablet wouldn't be always on? Can someone explain me? I think it doesn't make sense to call it "Always Connected" without notifications being able to pop in anytime (like Skype calls and messages). I thought this was the point of Always Connected PCs.
  • I agree - that doesn't make sense. Zac - can you clarify? In addition to above, if something like "Hey Cortana" isn't always listening they should not classify this as "Always Connected" - that will only confuse the market.
  • Always Connected PCs are simply PCs that can get access to internet via LTE. Not everywhere has WiFi, which is why LTE-enabled PCs are ACPC certified. There are already plenty of Intel ACPCs on the market.
  • These PCs do not have close to weeks of standby with "always-on" capability. You are promoting what Intel wants to make you believe. It is irony, that you are linking an Intel website, because naturally Intel would make you believe that ACPC are also possible with Intel SoCs - which they are not because they are far too inefficient. Could you also follow up on you certificate claim? Who is conducting the certification process based on what? From my information Intel just wants to water down the ACPC term - for obvious reasons - and there is not certification process at all. If you cannot win with outdated technology you try to win with marketing. As a journalist, you should severely criticize Intels practices instead of chiming in. It should also concern you that the Windows ecosystem is locked into ineffcient Intel SoCs while the world moves on - your articles do not help at all to get out of the lock.
  • The only reason is that Win ARM sucks!! big time..the whole pricey LTE is simply BS! Now fanbabies here will come and say that the envy is so expensive because it has LTE :)) and not because HP went full retard, AGAIN, hoping that some fool would actually pay for that.
  • lol, the irony 🙄
  • The HP is expensive because it is premium. Ask yourself how much does an iPhone or Galaxy phone cost full retail and you're in the ballpark. We associate Qualcomm processors with cheap, but they aren't at the high end and neither are phones. The HP is a high end device.
  • And we all saw how that went haven't we? Iphones and Galaxy phones SELL and people want them. I don't recall that junk X3 being sold very well...not because it was not a good looking phone, well built, but because it had a MEDIOCRE and DEAD OS at a price that no sane consumer or business user would had ever paid for.
    Same goes for this Envy 2in1. Good looking, but with crippled performance at $999 at base configuration...We'll talk more this Fall when the situation will be the significant sales and less than 1% usage of this device...
    You talk about students, high school kids...seriously?? Those teenagers want the cool stuff! and MS is not cool for them. They want decent performance and not being stuck with a DEAD JUNK app store that has no place on a DESKTOP OS, a junk tablet experience and buggy OS updates. But hey, I'm the bad guy for not taking the bait again...sorry I ain't that stupid and naive.
  • Hey MMGN, please refrain from using the "R" word please. I have a son with special needs, and I find that very very disrespectful and rude. Thanks.
  • Sorry for that, I'll rephrase: HP bought the same crap again from MS (brings back memories of Elite X3 super priced phone that no one bought)
  • 👍 Thank you much!
  • So if the budget non Pro Surface devices were not about performance than wouldn't this new device be perfect for an ARM processor? Should also be easy to disable LTE. I don't get it...
  • Personally I think, QUALCOMM should drop the LTE licensing cost for WoA and try to grab market share, but still include the LTE modem in all models. LTE should be standard on all mobile devices. If you don't use it, turn it off. simple.
  • I’m sorry, but this just doesn’t make any sense. First off, the Intel Core-Y series is a good chip but it is relatively expensive. The cheapest model is $273 when purchased in bulk. There is no way a device with that processor can go for under $500 (and if we are being honest probably not even $550). Look around for devices with those chips, and you’ll see what I am talking about. If MS wants to hit the $400 price point, they are going to have to use the mobile Celeron chips (these start at a little over $100). However, those are extremely “meh” when it comes to performance, battery life, etc. The Snapdragon chips seems like they would have been a good fit for this scenario. They are relatively cheap (rumors are Snapdragon 845’s run about $100) and the education market seems like it would be a good place to introduce the more performance-limited use cases of Windows on ARM. I doubt the whole modem cost has anything to do with the decision to not use Qualcomm chip’s either. LTE modems are built into all of them already so the only additional expense would be adding the LTE antenna and certifying these devices to operate on US LTE networks, which shouldn’t be an outrageous cost. What possibly happened was that Qualcomm requires WoA devices to heavily tout LTE in the marketing material. MS understands that schools and other education environments generally do not want LTE in the devices they issue (they will never buy plans b/c data is too expensive, LTE is unsecured, etc.) Ultimately, MS probably decided (maybe incorrectly?) that LTE shouldn’t be a main focal point of this new device, which forced them to go the Intel route. However, based on the Intel chip options I mentioned at the beginning of my comment, it seems like this Windows device may be poorly positioned to take on the $329 Ipad. A lot of this looks like it’s beyond Microsoft’s control, but I still don’t envy the choices they are having to make because there doesn’t appear to be an obviously “good” one in the bunch.
  • > ... education market seems like it would be a good place to introduce the more performance-limited use cases of Windows on ARM. Education market relies on a lot ox x86 software, which nobody is going to bother porting to ARM. From personal experience with biology and literature (!) majors, I can tell you that R Studio is the must in a lot of US schools. Engineering majors love their Matlab/Octave and I am sure, anyone with any ties to academia will list a few more things.
  • How does Chromebook compete so well then?
  • Does it? I don't seem to see all that many of them. Which college/major are you basing your assesment on?
  • It depends if "Education" market means selling to schools or selling to students. If selling to students (which I believe is the purpose, not schools -- they leave that largely to OEMs), the ability to play games well is important and the current implementation of WoA is not ready for that. This should change over the next 18 months, but if releasing this year, Intel is still a must for game performance across the lions share of Windows games.
  • I agree the idea that this is about price doesn't make sense if Microsoft puts a Y-series processor in there. However, I don't think Microsoft would choose the chip based on "marketing material" requirements. My guess is that it has to do with compatibility and the nascent state of WoA.
  • There's no way Intel's partners pay anything close the the stated prices on ARK. Case in point, the price of the i3-7100U is supposedly like $280 (same as the 7Y30). MSI sells a NUC-like barebones mini PC for $285, and Shuttle has one on sale at Amazon for $245 right now. No way they're paying more than $200 for those chips, and probably even less than that. Microsoft can probably hit the $400 price point with basically zero margin, because they're going to make $50 from every person who buys a keyboard cover to go with it. Also, having a compelling product for the education market is essential to the long-term health of the company, so it's worth breaking even if they're moving units.
  • I figured it would be about Windows 10 s vs full Windows . I'd hate to be locked to the Store
  • Greyfish, Now that iTunes is in the store, it's cool with me. I think I have probably 3 programs that are not from the store. It's not as bad as you think on a computer since the web browser is good. I just wish texture had a web browser experience to use, then I would ditch my ipad air 2.
  • hmm MS Devices VP Panos and his teams. Needs to put out products that sell. they wont sell well if they do not work well. an Intel Core M CPU computer for 400 Dollars will do basic Computer chores such as research Web Surfing, Netflix, Ebooks, Email, Spotify and Microsoft Office & it's a win win device for Students, Business People and Average people who do not need to do high level Video editing or Computer aided design work which require more powerful CPU's. This product should sell well to People who know it's limitations
  • Nah, don't be so pesimistic, I think Intel CPU's in 2018 have matured so much that compared to flagship CPU's of 3 years ago (for example Intel Core i7 5500U) which came on expensive Ultrabooks like XPS 13 are today miles away of 8th generation CPU's.
    Just as an example the Core M CPU on this budget 10 inch will get Octane benchmarks of more than 22000, and today my Intel Core i7 5500U with Spectre/Meltdown patches has slowed down about 30% so I think this will have great performance.
  • Atom to core m would be big improvement in performance specially if they use ssd with it. Compared to Windows on arm devices it would sacrifice instant on standby and battery life to keep software compatibility.if priced at 400 to 500 then it is winner. But Did intel launched new generation core m's or It will end up with last years chip?
  • Most likely what happened is Microsoft's support of Windows on Arm forced Intel to lower the price of Core M/Y which are outrageously priced the same as Core I ($280-330 per tray of 1000). Windows on Arm is a mild curiosity now but becomes very intriguing in 2019/20..... that's when the snapdragon 655 will be as fast as the snapdragon 835 with LTE for $40-50 (Atom pricing) and the snapdragon 855 will likely be 50% faster than the Snapgragon 835.
  • Don't believe Intel's pricing. You can buy a whole laptop with an i3 for less than what Intel says a mobile i3 costs. I'd say the actual cost is closer to half of the ARK price, and Intel makes of a lot of the difference by forcing their chipsets and Wi-Fi cards onto their partners. Partners can either buy the CPU or CPU/motherboard/Wi-Fi for the essentially same price, and that price is less than the price they show to the public.
  • ARM is too weak for a PC OS. Windows is a PC OS.
  • nah. Windows Os runs fine.
    x86 Emulation speed is problem.
    Snapdragon 835 is close to core i3/m3 with better gpu.
  • You have rather shallow understanding of what ARM is capable of and the Windows OS. Windows have no problem to be on ARM, it is designed to run on any architecture thanks to the NT Kernel. The OS itself now is optimized for ARM. The problem is that the classic desktop apps are only compiled for x86/x64 architecture, thus on ARM they have to be emulated, thus reduced performance. ARM is actually quite powerful CPU also that there are servers running it. Also Snapdragon 845 is comparable to i3. Still at the end, that depends on the software.
  • > Windows have no problem to be on ARM, it is designed to run on any architecture thanks to the NT Kernel. You are right, but... NT on MIPS/Power PC newer gained any traction and NT on Alpha was always a second class citizen. Also, WIndows NT 4 introduced some serious performance inprovements, which, unfortunately, relied on Intel-specific things.
  • You guys are completely forgetting AMD. AMD made some sort of collaboration with Qualcomm to include 4GLTE modems in their upcoming line up of mobile PC's. AMD Ryzen 3 Mobile APU with Vega Compute units gives more bang for the buck then anything Intel alone or Qualcomm alone has to offer. That is what I am really waiting for. The APU by itself is only 99.99. For f*cs sake Microsfot, and OEMS, go for red! Go for AMD!!
  • > AMD made some sort of collaboration AMD has been making vaporware for the last few years... and, given that they have *never* done good performance/thermal combination, the price is the only thing they have going for them.
  • Because games. Surface (non-Pro) are indeed targeted at students and more casual users. For this market, the ability to do some light gaming is important. The ARM chips are great for work, but bad for play on a Windows computer right now. Unless the Surface is going to be locked to the Store and there will be a lot more Store-based games in time for launch, this would ensure a bad experience for Surface customers. The solution is, of course, to go Intel for this market segment at this time.
  • Gaming? Yeah right. A Core M CPU can barely pull off League of Legends, and nobody is going to play LoL on a 10" screen. For casual games, you're better off with an iPad or Android anything. For PC games, a 10" tablet will suck.
  • Well... I can run GeForce Now on the fairly underpowered machiine and play Overwatch at 1080p/60fps thanks to rendering happening elsewhere. Do you want to guess how willing NVIDIA will be to "recompile for ARM64"?
  • Mike, that's exactly the point -- you'd be better off with an iPad or Android, but that's especially true if it can ONLY play store-based games. The CPU matters far less than the GPU, but in emulation, I doubt (confess I don't know this for sure) there is a good tie in yet for hardware acceleration of graphics functions. However, there are MANY THOUSANDS of older Windows-based 32-bit games that will play very nicely on any Intel chip, absolutely including a Core M. If it can be promoted reminding about the library of games and entertainment options for Windows, or support Thunderbolt for an even more powerful external GPU and connection when in the dorm room, that's a radically different sales pitch than, "Run MS Office, a few other Store apps, and get long battery life," which is how students would hear it without some gaming support. That guarantees they head for iPad or Android. This is definitely all speculation on my part. Personally, I hope they release it on an ARM chip and press harder for bringing games and other apps to the Store, but if they're going with an Intel chip, I think access to games is a key consideration.
  • Here's a little sampling of PC games that play via touch well Win32: Icewind dale enhanced edition, planescape torment enhanced edition, trine 2, the civilisation games, sims 3, the witcher 3, shadowrun returns games, legend of grimlock games, banner saga, calvino noir, UWP: the room games, the "go series", asphalt games, modern combat games, leo's fortune, Minecraft, tiny troopers, halo spartan games, Botanicula (and many other great UWP games). You can also run fruity loops, adobe illustrator in touch mode - try that on an ipad. As you can see because this includes titles you can't get on ipad,the ipad is not a clear winner in terms of casual gaming at all. The windows tablet, especially with keyboard certainly represents a better gaming option than an android device (that expands the selection of playable games to include other legacy titles like doom, and need for speed most wanted, which play on CPUs as weak as the atoms with enough ram).
    I'd definately much rather play need for speed, than asphlat. The other neat trick with windows tablets, is a stylus will make a whole range of mouse based games like top down rpgs and rts work, without needing a keyboard and mouse, because all you are missing is mouse over. The only problem really with casual friendly apps and games in windows, is they are not all in the same store, and you need to have an idea what you are looking for. But I don't at all agree that an ipad is better for casual games. Maybe very marginally, and mostly because all the games are in the same store. The list of actual quality titles on ipad is only slightly larger, and most of the really good ones, are in fact, PC ports with slightly adjusted interfaces. Given there are some decent UWP games, and some touch friendly win32s, with the capacity to play a massive backcatalogue of legacy games with the addition of a stylus or keyboard, and even use some power user apps in touch mode, or with stylus - it's really not fair to say the ipad is a better option. What I would say, is that some software is not suitable for smaller screens in touch mode, and the OS isn't quite as optimised yet - but that still leaves a large realm of quality software to sell it on a smaller screen. There are definately reasons why someone would go for a windows tablet, functionally. And that point of difference will only improve over time as the platform expands into the hybrid and tablet market (Windows is actually the only platform that's growing in tablets, ipad and android are both shrinking).
  • This said I have no idea how windows on arm manages those same legacy and light games. If the projection of 70 percent capacity is correct (once cached), and we can count ARM chips as 30 percent more powerful as an atom (which I am pretty sure we can), then I don't think this prohibits playing doom, or need for speed on a ARM chipset either. I think that most likely windows central are overstating the emulation factor, and probably largely because you need to run the app before it caches the converted binaries (ie the first time you run an app, it runs like a dog, and it doesn't run properly to the second or third time - essentially you'll need to "prime" it before you actually use the game or app, get the graphics engine loaded in the cache). Based on what I know of ARM chips, and the emulation - I think you should be able to play all these games, maybe save witcher 3 on WoA. But you'd have to load and run the game in a slow painful way, a few times before actually playing - and that is perhaps a good reason to avoid ARM for now, for the student market, as they do, as stated play "light games". Admittedly, the most common useage is youtube and Netflix, and women don't generally game, so it's not like there's a market within that wouldn't miss it, but to capture the whole market, you'd want light gaming ability. However I also don't agree that students have no use for mobile networking. People with laptops as their sole computer take them _everywhere_. On the bus, on the train, to freinds places. A students laptop is like their constant travelling companion. A ten inch model would be even moreso, especially if it was slim like an ARM tablet can be. So IMO humble opinion it's a bit of a toss up between mobility and battery life (and video consuming markets), and power and gaming for those markets, as far as students go. It would be kind of smart if possible to release both - a thin light mobility version, and a thicker workhorse.
  • If you are aiming for the educational market, you need good legacy support. As much as I like the W10/ARM/Polaris directions education will always be about ten years behind. Hell our projectors still have VGA inputs.