The eyes are the Windows
|
|
|
Intel claims Windows on ARM will require developers to rejig code
CHIPMAKER Intel expects software developers to have their work cut out for them when Windows 8 hits the ARM architecture.
Microsoft's decision to produce a version of its upcoming Windows 8 operating system for the ARM architecture has led some to predict the demise of Intel in the smartphone and tablet space. However Bill Leszinske, general manager for Intel Atom told The INQUIRER that "a lot of software work will need to be done" by application developers for Windows 8 on ARM.
Leszinske pushed Intel's ability to "rapidly modify chips for needs" and said that its x86 architecture will lead to a "faster time to market" as a big selling point for Atom. And contrary to the belief that ARM will be the only one to benefit from Windows 8, Leszinske believes that it will help Intel, saying "past platforms are a bad judge of the future".
Intel's big selling point against ARM is the fact that Intel's x86 architecture has for the most part been the primary home of Microsoft's Windows desktop operating system. The only other architecture Windows used was DEC's Alpha, but sadly Compaq and then HP put paid to that.
While developing for ARM cores shouldn't be too different, and frankly good developers won't need a long time to get to grips with Windows on ARM, that learning curve is one of the things Intel is hoping will stand it in good stead.
Leszinske said, "The investment to establish a new architecture is incredibly high and finding ways to manage to evolve software over time is the key to success. Someone could come up with the world's greatest new microarchicture today but that doesn't mean it will get adopted because there's no software or no developers."
ARM told The INQUIRER last month that Windows 8 will also help its push into the server market, which Intel dominates with its Xeon line of chips. Leszinske said that he expected a similar "Atom versus ARM battle" in the server market, mirroring that in the smartphone and tablet market.
In the conservative server market Intel's advantage over ARM is two-fold. Intel's extensive work with both Microsoft and Linux communities has served it well, and ARM will have to do a lot of work in order to develop the "software ecosystem" that Leszinske was banging on about to push its chips. Secondly, Intel's brand power in the server market is significantly higher than that of any ARM chip vendor, which is something that will take years, not months, to change.
For Intel it might just be software rather than hardware that saves the firm. µ
Tags: Intel
Share this:
Share
I was a software tester for Microsoft in the way back when that they were supporting the Alpha processor. The Alpha boxes were hugely expensive, extremely well-built. But awful.
One proble is that they had a really strange way to store the BIOS on the system hard drive, which made updates very difficult. Installing Windows on these boxes was usually many orders of magnitude more difficult than X86 machines, since they had a whole bunch of special drivers that required complicated insertions of floppy disks and hitting the right keys to answer various questions. You had to carefully follow a 20 item list. It sometimes took hours to get Windows to install on one of these. Once installed it usually worked OK, but it generally wasn't as stable as the X86 build. They were quite fast, but not not that much more than a good X86 box.
Somehow, I don't think the new ARM devices for windows will have these problems possibly because they will be inexpensive, portable devices, and as such will have to be user friendly. It cost DEC a lot of money to support such unfriendly machines, which may be why thet were eventually purchased by Compaq.
Either ARM is successfully emulating x86 code or there is going to be some unhappy customers from common consumer standpoint.
ARM running in an apple ipad isn't a big deal.
ARM running in PC based netbook is. People expect more flexibility in something that windows/linux based and if it's not there, people won't buy it.
It might be fine for the server marketplace. Depends on how dependent the servers are on the x86 or if its' emulated well.
It sounds like it could be one big fuster cluck to me though. This could easily be intel FUD, They do that too much.
Yes, you are pretty much the only one writing Windows Apps in assembly. Assembly is pretty much restricted to embedded and very old legacy systems. C++ code will probably only need a recompile between ARM and x86 and .NET requires no changes so it should make most of our lives easy. (.NET developer here too).
ARM cores can be configured to either little or big endian. I expect it will probably be little endian on Windows because it would make people's lives easier. Take it from this young punk.
Am I the only programmer here who works with data in memory directly on a byte-level, or bit-level? Is ARM little-endian or big-endian? Kids these days don't even know what I'm talking about..
So, Bill Leszinske, are you really suggesting that application developers refuse to support their customers who choose to buy ARM-based Windows 8 tablets, while supporting their customers who choose to buy Intel-based Windows 8 tablets? I'm pretty sure it's Microsoft who's forcing the decision, here--if there exists Windows 8 for ARM and Intel, my Windows apps will need to support both, lest I write off potential customers (just to save a few hours of effort???). Sorry, Intel, but /my/ customers really don't care about micro-architectures.
Itanium is not really part of the x86 family either. It had to emulate x86 code.
"The only other architecture Windows used was DEC's Alpha, but sadly Compaq and then HP put paid to that."
Er, no. Windows NT was developed on the MIPS platform, then ported to x86. This happened not because MIPS was more popular, but because insisting on x86 as a secondary target ensured portability.
Windows NT was also ported to IBM's POWER architecture, and I've got the CDs to prove it.
It is unlikely that there will be significant difficulty in compiling most source code for ARM; the major loss will be x86 legacy apps.
This is complete fud.
@Nick, I agree... for the most part mostly all .NET programs that follow the standard libraries should not really be affected.
But not all coding uses .NET just yet. Many open source projects will need to recode and some may be stuck if they have dependancies on other open source projects. But I don't think it will be all that bad. The Initial ARM machines will be tablest and MS will most likely have all their productivity software running already... and Adobe would be smart to make sure flash is running fine... by the time OEMs get serious regarding low power desktops that are powered by Quad core ARMs from the likes of Nvidia and others... developers would have had time to start porting the approprate software for the market segment. It is all about time management, and what MS offers in the way of beta on the ARM plat form. They may need to team up with a Manufacturer so developers can get started...
"Rejig" is an amusing term. Unless microsoft does a very half-assed port the only rejigging required should be a rebuild of the code for the target platform. The one major exception to this will be if the target has a vastly different memory profile which could result in major changes to heap management but that remains to be seen. I have always the RTL's manager and hopefully they do a decent job of porting that also.
Very good point Mr. Bentley. Apparently some have forgotten. Rumor has that there was a warehouse full of PowerPC machines with windoze on them that never made it market. I believe that was shortly after the divorce with IBM that resulted in OS/2.
It is interesting to read through some of the headers in the windoze SDK because you can see just how many processors have been supported over the years. There are lots of ifdefs for the various CPUs in them.
You're forgetting high level languages such as .net which a vast majority of developers are already skilled in. Windows phone 7 is dependant on it. It won't be down to developers using .net to make that change, Microsoft will already have altered their framework accordingly and code will port seamlessly.
You're forgetting high level languages such as .net which a vast majority of developers are already skilled in. Windows phone 7 is dependant on it. It won't be down to developers using .net to make that change, Microsoft will already have altered their framework accordingly and code will port seamlessly.
Besides the Alpha processor architecture Windows also ran on IBM's powerpc architecture.