But this is the main reason why Surface RT bombed in 2012: Microsoft sold it as a "Windows" tablet - the connotation being that users could run Steam and Chrome on it, not specially-compiled apps running in the anemic Windows Store App sandbox.
I appreciate things are different now in 2017 with x86 emulation (or dynamic translation) and the ability for most Win32 programs to be repackaged as *.appx - but Microsoft misses the implication to consumers again: users expect the freedom to install and download software from wherever - just like their x86 desktop machines. No-one is asking for an iOS style App Store for Windows and it's starkly visible by how poorly the Windows App Store is still doing.
Cynically, I suspect everyone at MSFT is well-aware of this and they fully expect Windows 10-for-ARM to sink for these exact reasons, they're just doing this as a repeat of their protest moves to remind Intel not to take their x86 monopoly for granted, and the addition of x86 support is a particularly bloody (but ultimately harmless) fist to the nose.
I'll point out that there is a very obvious use-case for Windows-on-ARM: low-power rackmountable/high-density server hardware for applications like web-servers and IO (not CPU)-bound database servers. No need for a 300W Intel Xeon box, or even a 150W Intel Core i7 - a 20W ARM chip will do. If Microsoft announced a no-shit, just-like-x86 SKU of Windows Server with no artificial restrictions with the same 10 year support period then I guarantee that Dell and SuperMicro would jump at building and shipping ARM servers - but this hasn't happened, instead we got a ridiculously impractical "Windows IOT" SKU which feels more like a toy than the basis for business critical infrastructure.
Server applications are much easier to port to ARM than desktop applications, given their lack of dependence on hardware associated with x86, like 3D graphics acceleration - or just users' expectation that a single executable binary should work everywhere. Servers are operated by businesses with professional sysadmins who know how to rebuild open-source applications for their machines or at least know the difference between downloading a program lablled "Thumb-2" vs "x64". Furthermore probably most server applications run on Java or .NET and can be trivially ported over to ARM.
The fact we haven't seen a true Windows Server ARM means that we can immediately dismiss consumer-level releases of desktop Windows for ARM for now.
> I'll point out that there is a very obvious use-case for Windows-on-ARM: low-power rackmountable/high-density server hardware for applications like web-servers and IO (not CPU)-bound database servers
...but why is this better than linux-on-arm? Windows server hasn't been generally very successful.
I appreciate things are different now in 2017 with x86 emulation (or dynamic translation) and the ability for most Win32 programs to be repackaged as *.appx - but Microsoft misses the implication to consumers again: users expect the freedom to install and download software from wherever - just like their x86 desktop machines. No-one is asking for an iOS style App Store for Windows and it's starkly visible by how poorly the Windows App Store is still doing.
Cynically, I suspect everyone at MSFT is well-aware of this and they fully expect Windows 10-for-ARM to sink for these exact reasons, they're just doing this as a repeat of their protest moves to remind Intel not to take their x86 monopoly for granted, and the addition of x86 support is a particularly bloody (but ultimately harmless) fist to the nose.
I'll point out that there is a very obvious use-case for Windows-on-ARM: low-power rackmountable/high-density server hardware for applications like web-servers and IO (not CPU)-bound database servers. No need for a 300W Intel Xeon box, or even a 150W Intel Core i7 - a 20W ARM chip will do. If Microsoft announced a no-shit, just-like-x86 SKU of Windows Server with no artificial restrictions with the same 10 year support period then I guarantee that Dell and SuperMicro would jump at building and shipping ARM servers - but this hasn't happened, instead we got a ridiculously impractical "Windows IOT" SKU which feels more like a toy than the basis for business critical infrastructure.
Server applications are much easier to port to ARM than desktop applications, given their lack of dependence on hardware associated with x86, like 3D graphics acceleration - or just users' expectation that a single executable binary should work everywhere. Servers are operated by businesses with professional sysadmins who know how to rebuild open-source applications for their machines or at least know the difference between downloading a program lablled "Thumb-2" vs "x64". Furthermore probably most server applications run on Java or .NET and can be trivially ported over to ARM.
The fact we haven't seen a true Windows Server ARM means that we can immediately dismiss consumer-level releases of desktop Windows for ARM for now.