Comparison of Surface Pro i7 with iPad pro A9X

[rbgaynor]

"Partly because of the licensing system QUALCOMM uses to charge, which apple is taking them to court over, that makes their chip more expensive on expensive devices, and cheaper on cheaper devices. Qualcomm have been able to get crème money from apple, whilst undercutting intel in cheaper devices. "

You're confusing Qualcomm's modem (radio) business with their ARM processor business. Apple and Qualcomm are arguing over royalties for the modem chips, Apple has nothing to do with Qualcomm's Arm processors (Snapdragon) as they have their own superior in-house ARM processor designs.

 

[wynand32]

If we're comparing the iPad Pro to a Surface Pro, then that implies its own limitations. That is, we can only compare tasks that can be done on both devices. So, we're not talking about running Adobe Premier or doing 3D renders in AutoCAD, because those can't be done on an iPad Pro, period. We must be talking about the same kinds of casual productivity apps (the vastly more limited Office Mobile, for example, as opposed to the full Office 2016 suite), watching video, playing casual games, and the like.

That "Surface Pro is 1.7X as fast as the iPad Pro" bit is aimed at people who will be choosing between those two devices. To many of those people, it means, which device is faster at running Candy Crush Saga, browsing the web, or checking email.

And I'd wager that Intel will be doing some of their own comparisons once we see Windows 10 machines running on the Snapdragon 835, because there's no way ARM will run Win32 apps even close to as fast as Intel Core processors will. Even if the Snapdragon 835 was as fast in general at running complex apps as an Intel Core processor (which I don't think it is), that emulation layer is definitely going to get in the way.

 

[Cruncher04]

because there's no way ARM will run Win32 apps even close to as fast as Intel Core processors will. Even if the Snapdragon 835 was as fast in general at running complex apps as an Intel Core processor (which I don't think it is), that emulation layer is definitely going to get in the way.

Of course emulated apps vs. non emulated would not be a fair comparison at all. However Win32 Apps are not inherently x86...they can be compiled for ARM. So the idea that Win32 apps need to be emulated is flawed. Only x86 Win32 apps need to be emulated...ARM Win32 apps do not.

 

[convergent]

Of course emulated apps vs. non emulated would not be a fair comparison at all. However Win32 Apps are not inherently x86...they can be compiled for ARM. So the idea that Win32 apps need to be emulated is flawed. Only x86 Win32 apps need to be emulated...ARM Win32 apps do not.

Do you think that developers are going to want to recompile for ARM? I see that being a problem if they need to have different versions of their apps for a different processor... will become fragmented and inconsistent.

 

[wynand32]

Of course emulated apps vs. non emulated would not be a fair comparison at all. However Win32 Apps are not inherently x86...they can be compiled for ARM. So the idea that Win32 apps need to be emulated is flawed. Only x86 Win32 apps need to be emulated...ARM Win32 apps do not.

Yes, of course, when I said "Win32" I was referring to the 16 million or so Windows desktop apps out there that the Windows 10 on ARM emulation layer is intended to support. The lack of support for these apps -- which I'm guessing developers won't be so quick to recompile for ARM, any quicker than they've been to make UWP versions -- is what killed Windows RT (among other things).

And, I would submit that even if Autodesk recompiled AutoCAD for ARM, it wouldn't perform as well on a Snapdragon 835 as it does a Core i7.

 

[convergent]

Yes, of course, when I said "Win32" I was referring to the 16 million or so Windows desktop apps out there that the Windows 10 on ARM emulation layer is intended to support. The lack of support for these apps -- which I'm guessing developers won't be so quick to recompile for ARM, any quicker than they've been to make UWP versions -- is what killed Windows RT (among other things).

And, I would submit that even if Autodesk recompiled AutoCAD for ARM, it wouldn't perform as well on a Snapdragon 835 as it does a Core i7.

I totally agree with you on that. There have been many platform change attempts on different systems that failed miserably. The one that I thing was very painful but did succeed was when Mac moved from PPC to Intel. But they were pretty much all in on it so it had to work. I have my doubts on whether moving Windows to another processor platform will succeed, and I guess its tied to Microsoft's UWP strategy. They need that to work for their long term strategy... remains to be seen if it will be.

 

[Cruncher04]

Do you think that developers are going to want to recompile for ARM? I see that being a problem if they need to have different versions of their apps for a different processor... will become fragmented and inconsistent.

From developer perspective, the app is the source not the binary so it is very much the same version and there is no inconsistency.
My expectation is, that open source Win32 apps will become available for ARM very fast, so are .Net/CLI apps.

And, I would submit that even if Autodesk recompiled AutoCAD for ARM, it wouldn't perform as well on a Snapdragon 835 as it does a Core i7.

Sure, currently if you want to run professional high end apps you better invest in an i7 anyway. I mean a Snapdragon based tablet (and so are Atom and lower end Core models) is not the best choice for professional applications anyway. Thing is, the newly introduced Cortex A75 is 20% faster than Cortex A73 in Snapdragon 835 and next years ARM will again be 20% faster. So ARM is closing the gap fast to high end Intel Cores.
By introducing Windows on ARM Microsoft lays the foundation for an ARM based Windows ecosystem before it is too late for them.