Opinions

How Full Windows 10 running on ARM CPU will affect computing future

So it finally happened! I have been predicting x86 emulation on ARM CPU ever since Intel stopped product development for Atom for mobile SoC.


@WinHEC 2016, Microsoft showed Windows 10 running on ARM

A few major things happened over the year to make this happen.

  1. Windows 10 Mobile Continuum
  2. Intel ending development of Atom SoC
  3. Qualcomm adding virtualisation support for SnapDragon SoC
  4. Various bridges for developers to easily port their apps to Windows 10
  5. Windows 10 Common Core (or why there’s no 64bit mobile Windows 10)
  6. Some special sauce from the brains @Xamarin
  7. Microsoft letting go of the last vestiges of Nokia

Let’s look at this point by point.

First point. Windows 10 Mobile Continuum feature has always been this dream of making your smartphone act more like a PC when connected to a large display + keyboard/mouse. However, people quickly notice far too many icons don’t work because the developer hasn’t written their app in Universal Windows Platform (UWP) format or their mobile app isn’t optimised for a larger display yet.

Rather than waiting for mobile developers to get around to adding UWP support, why not ask PC developers to scale their apps down? And at the same time, support legacy Windows program through the Desktop to UWP bridge so it’s possible for Windows 10 to run these UWP on ARM64 SoC.

Second point. (This is my speculation.) Intel has made a terrible mistake in ending Atom SoC development. They pushed Microsoft to seek an alternate SoC and they are the probably the reason Qualcomm added virtualisation support. I don’t remember Google/Android asking for x86 capability.

Third Point. Despite how fast Snapdragon (SD) SoCs are, the Atom X7 can keep up with it easily & in fact outperform it, albeit using slightly more power. The reason is every core of Atom can process SIMD (DSP) SSE instructions which gives it an advantage in terms of mathematical calculations. The letdown of Atom is in the GPU. Unlike Qualcomm, Intel obviously doesn’t want to put a flagship class GPU to compete with their Core CPU.

Recently, Qualcomm announced DSP & Virtualisation support for their next-gen SD835 SoC with an even faster GPU. This allows Microsoft to enable many of the desktop class features that required DSP to accelerate their functions. Many of Windows 10 Mobile functions like its User Shell & Cellular functions will be integrated into full Desktop Windows 10 as shown in the video. Finally, full Windows 10 can compete with IOS & Android for Tablets.

Fourth point. Microsoft is a software company at its core. Their oldest and strongest products are their OS and Visual Studio development environment. It comes as no surprise they can create conversion tools to directly take other platform codes and make it work on theirs. It’s also no surprise when you consider the fact that Windows NT (which Windows 10 is based on) was created to run on multiple CPU architectures.

Fifth point. For the longest time, Windows Phone fans had been wondering why Microsoft refused to create a 64-bit version of Windows 10 Mobile. I have always argued that mobile platforms don’t need 64-bit at all because no single mobile app will ever use 4GB of memory. Now we know the reason is because Microsoft has been concentrating in making full Windows 10 even more mobile.

We already know that Windows 10 shares the same kernel & some of the hardware stacks between PC, Xbox, Mobile, Hololens and IoT. However, what most people don’t realise is Microsoft has been breaking their Service Stacks into smaller and smaller pieces with every new revision of Windows 10.

Just look at your Services manager in Computer Management, there’re more and more Services running in the background. However, RAM usage hardly increase. The reason is to allow more services to be compartmentalised so Microsoft developers can pick the services relevant to that particular platform and reuse as much code as possible. This isn’t possible if the services are large and monolithic in nature.

My prediction is, instead of emulation, Windows 10 will run natively in ARM64 mode. Not surprising since Surface RT IS running an ARM version of Windows 8. This means all (desktop) Windows 10 services AND first-party apps will run at full speed since they’re compiled for ARM64.

All UWP apps can run in either native ARM or virtualised x86 mode depending on the developer. Also not surprising because when you run a 32-bit program in 64-bit Windows, Windows will launch the program in Windows on Windows 64-bit mode (WoW64). So now, instead of apps running in an x64 sandbox, it’s a Windows app running in an ARM64 sandbox.

Sixth point. I bet the brilliant minds at Xamarin had been cross-pollinating with the minds at Visual Studio to create the most powerful x86/x64/ARM cross-compiling toolkit possible! I’m guessing these are the people convincing Qualcomm to add virtualisation support for the latest SD. The power of software!!!

Seventh point. I’m guessing former Microsoft staff of formerly Nokia mobile division still have a static mindset about the future of mobile and smartphones. It’s probably good they’re let go to work on other worthwhile projects like the new Nokia-branded smartphones ODM by the China firm that bought over from Microsoft, or SailFish for the Russian bloc.

However, we also know that certain key ex-Nokian went to work at Microsoft Research. I wouldn’t be surprised if these are expert in camera and cellular tech.

The future. This part is my speculation about Microsoft’s strategy moving forward.

If Intel had continue to invest in Atom SoC, we’ll probably have the mythical “Surface Phone” by now that can run UWP & Legacy programs through Continuum.

Qualcomm seem to have fill that gap but I’m still skeptical over the emulation performance. However, if the Adobe Photoshop presented by Terry Myerson in the video is running in emulation mode and the video wasn’t edited, then I’m cautiously optimistic. The “Surface Phone” can now get back on track but don’t expect it to ship until end of 2017.

This is also the first time desktop Windows 10 gain cellular radio stack and mobile-context specific features related to voice calls and sms. The Skype Preview app is already hinting at the direction Windows 10 is taking so it’s more proof of mobile features added to Windows 10, rather than more desktop features added to Windows 10 mobile. Microsoft is moving in the opposite direction as IOS & Android.

With a Cellular stack built into Windows 10, Windows will finally become a full fledge mobile OS. With ARM’s more simple RISC architecture, power management should be easier to manage and Windows 10 should have much better standby time. Intel Atom (Enhanced SpeedStep) has great running time but the standby time still lacks far behind Qualcomm’s Snapdragon high/low-power cores. It’ll be interesting to note how Windows 10 handles the ‘big.LITTLE’ Kryo 28x Cores.

Windows 10 Continuum will also have a combined dual-shell or a morphable shell (called CShell) that switches interface depending on the screen (not device) context. Maybe now we’ll finally have a Landscape mobile Start Screen, and of course, multi-screen is naturally supported. It better support dock-able Windows for larger 5-7″ screens too!

Ultimately, the “Surface Mobile” can be your only computing devices. So what kind of a device is this Surface Mobile? Well I think it’ll be running a SD835 clocked at 1.9Ghz on battery, and 2.45Ghz when plugged into the Microsoft Display Dock. It should have 4GB of RAM and has a 5.5″ – 6″ display to allow adequate space for heat spreading. A 1440P display is good enough, a 4K display will be a waste of battery and GPU cycles. Now, Windows 10 requires a lot of space, so 64GB of eMMC 5.1 Storage is required.

Where does that leave Intel or even AMD? Well, Intel has abandoned Atom because they have probably figured out how to make a Core series based SoC hit the 2W TDP but their 10nm fab is having yield issues, which is why CannonLake is postponed. AMD has also announced new APU that can step through the clock at 25Mhz increments thus providing more aggressive power management but being fab on a 28nm process, I reckon power consumption will be higher than KabyLake.

What this means is, Intel is trying to make their Core m CPU fit into a mobile environment so the CPU wouldn’t fry your smartphone. And AMD new Ryzen CPU isn’t ready for mobile.

Qualcomm just released information on their latest Snapdragon 835 SoC.

Based on the information released, SD835 supports new 3D audio processing and enhanced camera image processor. It looks like a future Microsoft Hololens 2 could be powered by the SD835! The Surface 4 could also be based on a SD835!

In the meantime, you can check up my past predictions which has come true below.

How Windows Phone 8 & Windows 8 can share software

OEMs can’t do it, so Microsoft DID IT!

Samsung losing patent case is Good for consumers!

Update 03 Feb 2017: Updated with the product codenames for future Windows features I have corrected predicted, like CSHELL and clearer explanation of CPU terminology.

Advertisement
Opinions

What Microsoft should bring to Windows Phone 7.8

Introduction

The bloggers & forum goers & tech journalists are still arguing over Microsoft’s announcement of all existing WP7 smartphones only being upgradable to Windows Phone 7.8 with only the new Start Screen as the only addition + that’s the end of the line for the WP7 line. Some have made-up their mind that’s the case, whether they’re fans of Windows Phone or not.

My opinion is more optimistic mixed with being realistic. But this article is not about that argument anymore. I’ve always argued that Microsoft hasn’t shown us ANY consumer features in the coming WP8 & by that line of reasoning, WP7.8.

So let us see where this line of argument takes us…

End of WP7? Maybe not…

According to this post from WPCentral, Senior Product Manager and Windows Phone team member Larry Lieberman has just given some indication that WP7.8 is not all it appears to be, just an additional Start Screen.

Towards the end of this video, Larry was evasive when asked if that’s all developers can expect of WP7.8, a new Start Screen. His words were carefully considered but it seems to indicate there’s more to come for consumers.

Greg Sullivan, Senior Product manager for Windows Phone has also told Pocket-lint, Microsoft has only shown a small portion of the consumer features at WP8 Summit in Jun 2012.

Let’s go back a few months & look at these news.

http://www.intomobile.com/2012/06/14/microsoft-helps-floundering-nokia-lower-prices-and-compete-against-android/

It’s speculated that these low-cost devices will run WP8 but if that’s the case, Microsoft can bring WP8 to all WP7 devices! Wouldn’t it be faster & cheaper to continue WP7.8 on single-core Qualcomm S2? So far, we’re hearing WP8 runs on Snapdragon S4 only.

Greg Sullivan had also confirm what many bloggers have suspected, that MS will not bring WP8 to WP7 due to time & manpower cost. It’s technically possible but it simply wouldn’t make it a great experience on current phones.

If Microsoft maintains WP7 & WP8, they owe it to WP7 users to ensure comparable features

Below is the list of stuff that should appear in WP7.8 that’s not hardware dependent:

  1. Screen capture (It’s possible because it’s already in most Custom ROMs)
  2. Custom sound for Notifications
  3. Support more Bluetooth profile like Contact transfer & file transfer
  4. User information backup/sync to Skydrive. Stuff like APP states (game progress), Message Thread, Video, Music, Podcast
  5. Skype Integration (Change its P2P operations to semi-P2P, super-node reside on servers rather than phones)
  6. Notification Center ala iOS 5 by swiping left or something
  7. A more full featured Network stack (OK, this one may be a stretch)
  8. Maps by Navteq with Nokia Drive
  9. Xbox Music (Everywhere)
  10. SmartGlass (if it doesn’t rely on DirectX)
  11. Graphics equalizer!
  12. IE10 (even if it’s without the Smartscreen feature)
  13. Like/Edit/Delete Comments/Pictures/whatever in Me/People Hub

What wouldn’t make the cut?

  1. Multi-threading (apps running in background)
  2. Integrated Skype (if they don’t get off their butt & re-engineer Skype network or do workaround in the client)
  3. Wallet (it’s only useful with NFC)
  4. DirectX for high performance gaming engine like Havok (Adreno 205 doesn’t cut it)
  5. Full IPv4/IPv6 network stack with VPN, IPSec, etc.

Microsoft will reveal all in the (rumoured) BUILD event coming end July, beginning August October, beginning November when they launch the new SDK for WP8.

What do you think? Comment below!

Opinions

What a Windows 8 Tablet should be…

* Follow-up of this article here “OEMs can’t do it, so Microsoft DID IT!”*

This article is about what I think a Windows 8 Tablet should be like.

A bit of background… My first Tablet PC was an Acer TravelMate C110 running Windows XP Tablet Edition. My current is a Fujitsu Lifebook T4210 (upgraded with Bluetooth, 4GB RAM, T7200 CPU & Self-encrypting SSD). The current machine started life with WinXP Tablet as well & was dual-booted with Vista, & finally replaced with Windows 7 Pro 32bit. SO I have a total of 8 years+ of Tablet usage.

I’m looking for a new Windows Tablet because my current doesn’t have a Touchscreen & it’s not something I can upgrade.

Based on my years of using Tablets, I have come to several conclusions which I’ll share here.

The new Tablet OS, Windows 8

Windows 8 is a rethinking of how people will use Tablets to Create Content in the future, versus, Consuming Content on the iPad & Android tablets. It’s made to run special Windows 8 applications + run your good old MS Office & Photoshop, etc.

By supporting 2 different mindsets, Content Consumption + Creation, Microsoft is trying to allow people who like iPads to enjoy ease-of-use & good battery life anywhere anytime. They are also trying to appeal to the office worker who wants a fast, stable & productive platform to get on with work to make the money to enjoy life. There is actually a 3rd mindset which isn’t well-defined yet, that of the gamer/home entertainment with XBox Live integration so we should have to wait to see how that works out.

Microsoft new vision is this. While on the move or onthe sofa, you’d be carrying a light-weight tablet that runs low-power content consumption apps from Microsoft Store & enjoy long battery life. When in the Office, you can plug it into a dock that’s connected to your Keyboard/Mouse/20″ monitor to work & run your old MS Office & Photoshop, etc.

After you’re done, you can bring the tablet home & plug it into your Home Entertainment center where you can play your games or watch streaming video on your 55″ 3D LED TV. That’s the vision Microsoft is pushing for Windows 8. Can it work? Yes! All the technology needed are there, WiDi, DLNA, 4G, HDMI…

Now Windows 8 Beta is coming out soon & should be almost feature complete. It should be interesting for people who has Touchscreen Tablets/Slate because they get to experience this vision first-hand. Some of the suggestions from Developers have been adopted in the Beta & it’ll be interesting to see how far MS goes to satisfy End-users when the Release Candidate comes out by mid-2012.

Hardware, the physical tablet must feel good & look good!

Many manufacturers have also pledged to launch Win8 Slates in H2 2012 after their Android Tablets fail to make a dent in iPad sales. If these new Slates are less than SGD$1000, it will attract buyers (with enough advertising & education.)

The current batch of Windows tablets (in Singapore), Acer W500, Asus B121 & Fujitsu Q550 all have fatal flaws that makes them unattractive to buyers. Chief among them is Windows 7 of course. There’re too much bad press about Windows 7 being a lousy tablet OS. Some of the complaints are true but many the reviewers don’t know what they’re talking about. Let me say again, Windows 7 is a competent Tablet OS when configured right.

It’s easy to configure Windows 7 to be easy to navigate using Touch. All the settings are inside the Control Panel (bigger icons/text & space in between/super large themes). StartBar can be docked to the side & gesture controls can be used to navigate the interface.

The problem, of course, is these things should be pre-configured by the Tablet makers themselves, not the end-users. Slapping a custom layer over Windows 7 desktop isn’t the answer as well! Therefore, the fault lies in these makers, Acer, Asus & even Fujitsu who just slap Windows 7 in as if they’re selling Desktops. No customisations whatsoever!

On top of that, these tablets are either too heavy or too bland or too slow. Take a look at the new Nokia Lumia 800 & 900. People naturally gravitate towards the better looking device even if the specs aren’t as fantastic as a dull looking but super-fast one! Early iPhones & iPads are the BEST examples!

The brains behind the operations, the CPU

The other flaw is CPU. Intel Atom CPU + motherboard + chipset was too expensive for its performance. Being single-core with low bus speed + a GPU that’s slower than molasses. It can’t even play a decent 720p video without dropping most of the frames.

The coming generation of Dual-core Atom CPU (Cedar Trail) should finally solve the problem unless Intel screws up somehow. Dual-core, higher bus speed, support for 4GB RAM + a PowerVR GPU should allow 720p video playback (clear enough for a 10-12″ screen or a 55″ TV). A typical person using this tablet for media consumption & business usage should have no complains about lag. Gamers wouldn’t be interested in Atom CPU of course but home entertainment is definitely doable!

Makers of tablet (Asus) also shouldn’t be using Core i5 in their Tablets. It adds weight & $ cost for heat dissipation & reduce battery life, or they have to use a bigger battery thus increasing weight. Considering Core i5 + chipset uses >30W while Atom + chipset uses <10W. A 4-cell battery powering an Atom N2800 should last >4 hours of actual usage with WiFi on & screen at half-brightness.

SSD prices have come down by a lot on the lower-end, 32-64GB. If storage is not enough, a 32GB SDcard or MicroSD is also dirt cheap. With all these drop in prices, I do not see why a Win8 Slate should be more than $1000.

Let’s guesstimate the costs of the perfect Windows 8 tablet…

MS Windows 8 Home OEM license – $130
Intel Atom N2800 CPU – USD47
4GB DDR2/3-800 SODIMM RAM – $50
Kingston SSDNow Self-encrypting 32GB – $100
IPS 10″ LED screen – est. $150
Li-ion 4-cell battery – est. $100
Motherboard + WiFi + Bluetooth – est. $80
Chassis + Gorilla glass + Touchscreen + Active Digitizer – est. $120

Total around $800 including box + packaging! Maybe I’m optimistic but I think $1000 for this Tablet is entirely possible! If you include branding/advertising, it’s still an additional $100 at most. Still near to the price of a White iPad 2 32GB. Tablet makers can even throw in a 32GB SDcard to act like a secondary storage & only add another $20 to BOM cost.

We can see from the Android camp that users are demanding higher & higher performance by using Quad-core CPUs & integrated high-performance GPU to save power & improve Android’s laggy nature. The irony is, Windows 7 is super optimised already & Intel is simply lagging behind on the low-power front. Nobody I know liked their Netbook due to the many (artificial) limitations that Intel put on the Atom CPU. The only decent Atom is the Cedar Trail CPU that I just mentioned but the price is not low either.

Intel should watch out since Windows 8 will be able to run on some ARM-based platform like Nvidia Tegra & Qualcomm Snapdragon S4. I’m not mentioning ARM-based Windows 8 because older Wintel apps like MS Office cannot run on ARM-based Windows 8 unless Microsoft emulate x86 platform on ARM which will be a major engineering undertaking, not to mention the high licensing fees they have to pay Intel. I may be wrong but I’m pretty sure MS will not do this for many years to come.

Conclusions

So the question is, why aren’t PC makers chunning out great tablets? Maybe it’s because there wasn’t a strong enough vision until Apple launched the iPad or Microsoft was so pre-occupied with the Anti-Trust lawsuits that they didn’t listen to customer demands for the past 10 years. Or even maybe because Bill Gates has left the building.

We have been pushing for lighter tablets with better battery life for the longest time but while Sony & Lenovo have 1KG laptops with >8 hour battery, they didn’t translate these to the Tablet form factor. It has been extremely fustrating for Tablet users like me for the longest time, but finally it seems that our prayers are being answered with Windows 8.

In conclusion, 4 things must come together for the Windows 8 Tablet to be successful or better yet, be desirable.

1. Windows 8 is already super efficient in terms of CPU/Memory performance. It’s the Interface that’s the biggest bet for Microsoft. So far the Metro design language seems to be winning praise from Press & developers. Windows 8 has 2 interfaces, Metro for media consumption & traditional for media creation. If Microsoft listens to End-users during the Beta & makes the necessary adjustment, Windows 8 Metro could be wildly successful. Many people are already praising the Metro interface on the Xbox 360 & Windows Phone 7 Mango, so chances are good.

2. Hardware makers MUST make their tablet shine! Nokia has shown that with a beautiful device like the Lumia 800 & 900 + enough advertising, people will WANT to get it! Apple basically made a name for themselves around simple beautiful design with tons of advertising. That’s why there are calls for Nokia to make a Win8 tablet based on that polycarbonate shell! I certainly would love a Lumia tablet but that HP Slate is cool as well!

3. Intel must support Windows 8 by making a CPU/chipset that’s fast enough & cheap enough. So far the Intel Atom CPU has been a joke in the market. More than anything, users were not satisfied with Netbook due to lag. It’s ok for the processing to be slow but it’s NOT ok for the mouse to stop working after clicking something. This is the problem with a single-core CPU on a desktop OS.

4. Price. Fujitsu Stylistic has been the pinnacle of TabletPC in the Slate format for many years, but few companies & even fewer individuals can afford the $5000-6000 price tag. So long as manufacturers stick to Intel Atom Cedar Trail, they can afford to use cheaper components. These few years has battled the economy of many countries. Having a fast enough tablet at an affordable price will be a major factor in determining how many buyers they can attract which will further drive adoption rates.

To say that I’m excited about new Win8 product launches later this year is an understatement, especially with all the cool Ultrabooks that these same makers can make. If they (Microsoft/Intel/Manufacturers) get the above 4 points right, we could get a iPad killer. Until then, we can wish & pray.