Slashdot Mirror


Embedded Systems Study Rebutted

Gabba writes "LinuxDevices.com has a rebuttal to the Microsoft-funded report purporting to show Windows nearly 4X more efficient than Linux for developing embedded systems. The rebuttal shows the study to be full of flaws in both design and execution."

28 of 203 comments (clear)

  1. The magic of studies by s20451 · · Score: 3, Insightful

    This is a tried and true method in the scientific community: if you disagree with the conclusions of a study, you can always call the methodology "flawed". That way, you never have to pay attention to results that are different from what you believe.

    --
    Toronto-area transit rider? Rate your ride.
    1. Re:The magic of studies by Deusy · · Score: 3, Insightful

      This is a tried and true method in the corporate community: if there is strong competition in the market place, you can always sponsor a ttuday to call it "flawed". That way, you never have to pay attention to the flaws in your own software that you refuse to believe.

      I'm sure this study, if impartial, would be by-enlarge accepted and address the flaws highlighted by said study.

      But how come the _majority_ of "studies" that debunk Microsoft's competitors are usually sponsored by the software giant?

      --

      Free Gamer - Free games list and commentary

    2. Re:The magic of studies by wik · · Score: 5, Funny

      > by-enlarge

      You've been reading too many spams. The correct expression is "by and large". :-)

      --
      / \
      \ / ASCII ribbon campaign for peace
      x
      / \
    3. Re:The magic of studies by pstemari · · Score: 5, Insightful

      Nice troll, but it doesn't fly. There are in fact objective criteria for evalating experimental and statistical methodology, even though both coal companies and the "alternative medicine" crowd wish it weren't so.

    4. Re:The magic of studies by LinuxInDallas · · Score: 4, Insightful

      Because they have the money to fund the studies. Now, whether or not that implies the studies are biased I can't say because I don't know the authors of these studies.

      BTW, if Microsoft funded a study that concluded that Windows wasn't superior for a certain task do you think they would allow it to be published? I think not. Because of that all Microsoft funded studies (that we read) will always conclude Windows is the best.

    5. Re:The magic of studies by jbottero · · Score: 5, Insightful

      What's the name of that new Linux guy at M$? This is the kind of stuff he needs to address. Microsoft can NOT now afford to be spewing out "whitpapers" that can go right to the shreader.

      Look, there is no "brain drain" at Microsoft, they DO have some smart people working for them. So why do they produce frilly lady fluff like this? If Microsoft wants to survive, they will need to work ALONG SIDE linux, have a reallistic approach to compeating with Linux, not just shitting out paper after paper that even marginal techies like me laugh at.

    6. Re:The magic of studies by fermion · · Score: 5, Insightful
      Perhaps, but in this case the methodology is flawed, and one needs to read no further than the first several paragraphs to see this.

      As clearly point out in the rebuttal, this was a statistical study and as such must follow precise rules to considered valid. First, any sample must be random. The study was not random but self-selective as the names were drawn from a web site registry. Second, the poll question must be made public to insure that they were not leading. Without these two criteria met, we must assume that survey is flawed.

      The problems mount when one reads that the data analysis methodology is not given. We can in fact live without know the specific data, but without the methods we must assume that the analyst crunched the data until they discovered the answer they wanted. This is a classic method of lying with statistics.

      The nail in the coffin, even if we assume the analysts are honest in all other respects, is that we do not know how the result were normalized. All we know is that for the projects developed by the Linux people cost four times as much as the projects developed by the windows developers. Were the Linux projects 4 times as hard? Did the Linux projects return four times the revenue? Were the Linux projects more costly research based projects while the Windows projects merely applications of work already completed? Were the differences in the way the various companies costed the projects.

      All the other stuff is just in the rebuttal is just a rational of why Linux development is probably just as good as Windows development. The fact is that the study has all the classic signs of a mercenary statistician massaging data to generate a predetermined answer. People wonder why kids are so stupid now. It is because companies like MS want to keep them stupid so they will believe these bogus studies.

      --
      "She's a scientist and a lesbian. She's not going to let it slide." Orphan Black
  2. This sort of thing is nice... by JessLeah · · Score: 4, Insightful

    ...but needs PUBLICIZING. As in, to non-geeks. Specifically, pointy-haired bosses.

    Wanna bet the pro-MS article will be the one most PHBs will have come across their desk?

    1. Re:This sort of thing is nice... by _Sharp'r_ · · Score: 3, Insightful

      Not sure how anyone mistook the original marketing material referred to as a "study" anyway. All you'd have to do is read it to see that it wasn't a real study in the first place.

      --
      The party of stupid and the party of evil get together and do something both stupid and evil, then call it bipartisan.
  3. Wow.. by nother_nix_hacker · · Score: 4, Funny
    Windows nearly 4X more efficient than Linux for developing embedded systems
    Windows and Linux can develop their own embedded systems?
  4. Shock! by Chicane-UK · · Score: 5, Insightful

    The rebuttal shows the study to be full of flaws in both design and execution.

    I'm not being funny here, but since when did anyone ever pay attention to Microsoft backed or funded tests such as this? They'd hardly be backing this comparative benchmarks and reviews only for their product to get slated. Every time I read a 'Windows 2003 Server is 2.3x cheaper than Linux!' type story (where they end up comparing to something like Solaris which.. duh.. isn't Linux!) it just bugs the heck out of me.

    I'd much rather cast my attention to impartial, un-biased sites such as /. for fair reviews of items provided by stores they have no relation to.. now where can I order a Zero Blaster again? :D

    And yes, I am just joking around before I get flamed to hell and back ;)

    --
    "Hey! Unless this is a nude love-in, get the hell off my property!!"
    1. Re:Shock! by portnux · · Score: 3, Funny

      The report was full of "flaws in both design and execution"? Maybe it was done by the same teams that ordinarily are involved with their products?

  5. maybe 4x more efficient but by joeflies · · Score: 5, Insightful

    after seeing ozzy trying out the iDrive on his BMW, I would guess it may not be more usable or more reliable

    1. Re:maybe 4x more efficient but by CausticWindow · · Score: 4, Funny

      I'm not quite sure what the iDrive is, but I read a story about the malaysian financial minister. He was going to a meeting in a brand new BMW, with computer driven everything. Suddenly, the embedded computer crashed which caused the doors to lock without a way to open them. Same for the windows and the hole in the roof.

      As it was 200+ degrees fahrenheit, he was almost cooked before a maintenance man managed to smash the bullet proof windows with a sledge.

      --
      How small a thought it takes to fill a whole life
  6. Is that your final answer? by segment · · Score: 4, Insightful
    Microsoft's "Windows Embedded" operating system platforms (specifically, Windows CE .NET and Windows XP Embedded) are completed 43% faster and at 68% lower cost, on average, compared with similar projects using Embedded Linux.

    Windows XP = money Linux = free to download

    I don't see where the money argument comes into play here? Before someone says something about TCO let me point out a humorous but true text on how Microsoft actually kills (link is a google cache).

    The report includes data from a survey of 100 manufacturers using 32-bit processors in a range of embedded projects and applications -- 50 using various implementations of embedded Linux, and 50 using Microsoft's Windows Embedded platforms(Windows CE .NET and Windows XP Embedded).

    Rubbish rubbish and more rubbish. They shouldn't have been so biased with the study. Which manufacturers were used? Give it a rest now MS. It's obvious for one if MS funded the study, it's bound to be swayed, however if they didn't fund the study, depending on the vendors, it's still bound to be swayed. Remember MS violated antitrust forcing companies to go MS or go to bankruptcy court. How is one supposed to believe any studies they'd do?

    I'm sure someone else is going to post a very good thorough post but we all know this is nothing more than utter bs.

  7. Define Embedded by spoonist · · Score: 4, Interesting

    First, let me be honest. I just skimmed the LinuxDevices article and didn't read the Microsoft article.

    One thing I've noticed among PHBs is an ever-broadening definition of "embedded systems". I've seen more than one project go down the road of using a cPCI system running Windows NT 3.51 (yes these are current systems running this old version) on a harddrive. These systems are calling themselves "embedded".

    This has been especially in systems that had serious size, weight, and power needs. Had I designed the system, I guess I would've used something like QNX or Linux on a much smaller processor, compact flash card, etc.

    I guess my point is that these days it seems like general-purpose computers are being called "embedded" when I see embedded as much, much smaller (e.g. no moving parts, a microcontroller, etc...).

    I dunno, I'm rambling...

  8. Obvious implications by serial+frame · · Score: 3, Interesting

    On a consumer level, there are no real benefits to using Windows on embedded controllers, or even developing for evaluation boards.

    Although I deviate from Linux in this example, it is still relevant to all open source embedded solutions. A few questions: is there an implementation of Remote Terminal Services for embedded versions of Windows, for easy manipulation of the embedded device in question? If so, what sort of licensing costs are implied?

    As demonstrated numerous times before, open standards such as VNC are superior in the aspects of platform-ubiquity, openness, freeness, and simplicity. A shining example of what would be a costly, if implemented, solution, under Windows would be the Ethernet board running Contiki.

    Oh yeah, and how many simultaneous threads, per-process-threads, and processes, do embedded Windows products support?

    One must also compare the existing products that can be compiled between embedded Linux and Windows. I'm willing to bet software written with POSIX in mind beats Windows.

    Excuse me if these speculations seem a bit armchair.

    --

    -
    And the Angel said unto me, "These are the cries of the carrots! The cries of the carrots!"
  9. crap for crap by dirk · · Score: 4, Insightful

    I hadn't seen the original report, but as soon as I saw it was funded by MS, I assumed it was marketting fluff and moved on. Now I see that the "rebuttal" is on LinuxDevices.com, and was meant to specifically prove the MS report wrong and that Linux is better. That means it's more marketting fluff so we can move on.

    When will people relaize that MS is not the only people putting out biased reports. I put the same faith in a "Linux is great" report by a Linux group as I do in a "Windows is great" report by MS.

    --

    "Information wants to be expensive" - Stewart Brand, the same guy who said "Information wants to be free"
    1. Re:crap for crap by barawn · · Score: 4, Insightful
      When will people relaize that MS is not the only people putting out biased reports. I put the same faith in a "Linux is great" report by a Linux group as I do in a "Windows is great" report by MS.


      They'll realize this when other people actually read the article in question.

      Go ahead. Read it. Carefully. Note that nowhere in the report does it say "Linux is great" or "Linux is better".

      In fact, to quote the article,

      Consider that in most embedded software development efforts, only a small portion of time is spent on platform issues. In virtually every project I've been associated with over many years, one engineer has selected the development environment, brought it up on the target hardware, and introduced the other engineers to its use. From that point on, everyone involved is focused on the application rather than the environment. Platform issues constitute only a small proportion of the effort expended on all but the simplest cookie-cutter devices.


      Note what's said there - it doesn't really matter what platform you develop on. That's what's said there. Develop on whatever platform suits your needs.

      Still think it says Linux is better?

      It could be, as John Lettice has pointed out, that developers of larger, more innovative, products tend to choose Linux because of the control and flexibility it offers, while developers of cookie-cutter devices tend to choose Windows because of the help it provides.


      What's that? There, they're saying "Windows is better for some things!" Blasphemy? No - they are actually trying to be unbiased.

      Granted. They can't be completely unbiased. But they're trying, which is the difference.
  10. Re:Yeah... by Voivod · · Score: 4, Informative

    Not to support Microsoft or anything, but your Dad isn't using "embedded" Windows. This is like complaining that RedHat 9 takes too long to boot on your i486 embedded CPU.

    Both Windows CE and XP Embedded are designed to let you remove whatever components you want. You can strip out the GUI, networking, swap files, etc. Windows CE can definitially be customized to boot in just a few seconds.

    On the other hand, as an embedded developer I must say that Windows CE is the WORST OS by far I have ever had to work with. It's so bad my company discarded 3 months of work on drivers and a BSP (Board Support Package) for our hardware because neither we nor any of our customers could figure out how to use it reliably. It's an absolute nightmare.

    Linux is very nice for embedded systems and I'd guess 40% of our customers are using it with our hardware, losing out to DOS believe it or not. The only OS I think is better for embedding is QNX. If you can afford it, QNX absolutely rocks.

  11. It's all about placement by danaris · · Score: 4, Insightful

    I didn't catch the original, and I don't know where it was publicized or anything, but I imagine it was in places that make it look legit. More importantly, it was in places that people who don't already use Linux might be likely to look. Rebuttals like this are nearly useless unless they are prominently placed in some widely read medium--and by "widely read" I mean someplace that at least an average techie, even if entirely M$-biased, would be likely to at least see the headline. I admit that I have no knowledge at all of the world of journalism, online or otherwise, but I think that people who write studies/rebuttals/articles/etc like this, showing up Microsoft and their precious status quo, should make significant efforts to get them in mainstream media.

    Unless someone already likes Linux, they're not likely to frequent LinuxDevices.com. Someone who already likes Linux is not the target audience for such journalism, or shouldn't be. We need to target it at the others, the people who don't like Linux, because it's articles like this that might make them like it, and it's studies like the one it's rebutting that make them not like it.

    Dan Aris

    --
    Fun. Free. Online. RPG. BattleMaster.
  12. Re:Yeah... by allanj · · Score: 5, Insightful

    One of the problems with windows is the inability to strip out stuff you don't need.


    I don't know what Windows version you're talking about here, but it most certainly is NOT recent Windows XP Embedded (as in Embedded XP SP1 - I assume you're sufficiently on-topic to be talking about embedded Windows stuff). You can mix and match your own XP configuration, and have it contain just the pieces you need.

    Don't need no steeenkin' GUI? Unselect it.

    Don't need no steeenkin' webserver? Unselect it.

    Don't need no steeenkin' Ethernet support? Unselect it.

    The list could go on for miles - it really works quite well. There's a target designer tool where just about anything is optional, and the database for that tool is HUGE. Module inter-dependencies are handled automatically, and the size of the target image easily available. You could argue that MS would need that tool, since no-one can figure out how their dependencies work anyway (and I'd agree with you :-), but that's besides the point. The point is that it's SOOOO easy to select/unselect options.


    And no, I'm not affiliated with MS in any way - I've just actually USED Embedded XP, and that brings a somewhat different view on things.

    --
    Black holes are where God divided by zero
  13. Extra! Extra! Read all about it! by brooks_talley · · Score: 3, Funny

    Microsoft says Windows better than Linux! Linux advocates disagree!

    In other news, a communist revolution has shaken Russia, and Napoleon suffered a shocking defeat at Waterloo.

    Cheers
    -b

  14. Use gnu/hurd for emedding instead. by Anonymous Coward · · Score: 3, Interesting

    At our company we use GNU/hurd on our embedded operating system. Everybody laughs about how gnu/hurd sucks but its perfect for embedded stuff. We produce Small kde-based web terminals for cyber cafe. On a Yontrix 386 compatible processor with 2 Mb of ram, it boots off a 1.44M flash rom. The KDE is a stripped down version of kde 1.4 with just Konqueror, kedit, and kicker installed. No Xfree86, a EGA frame buffer is used. If you thought KDE was bloated, just see our web termials . Since hurd is completly free and easy to code for, it took us only two hours of programming time to make this system compared to 3 weeks for a linux soloution..

  15. This Study *is* Flawed by 0x0d0a · · Score: 4, Interesting

    Okay, a lot of pro-Linux studies have their own problems (frankly, I don't put much stock in "studies" any more, especially vendor-funded ones).

    However, the numbers this one used are *ridiculous*. Total Cost of Development?

    Okay, let's see. Of the parties surveyed, cancelled Linux projects cost more than CE projects. This Jerry guy (he's got a PhD, so he must just be dishonest, not stupid) then uses this as a basis to claim that Linux is more expensive than CE. He's got to be kidding me.

    By the same metric, all C++ software should be replaced by bash equivalents. Why? Because the average cost of a bash-based project is much, much smaller than the average cost of a C++ project. Of course, there's the little additional detail that the sort of projects one uses bash on are much, much smaller and simpler. That is, of course, the factor that makes the huge difference. However, you can conveniently ignore that tidbit.

    Somebody tried to do the same study with Windows and some Sun servers back in the day to show that Windows made a far cheaper server. Well...yes, but most of the servers being used to average out Windows cost in the study were small, departmental servers that nobody was spending much on. The Sun servers were the far more powerful and capable systems for things like eBay's back end that had technicians swarming all over 'em. Sure enough, the Windows boxes had a lower average maintenance cost.

    Average total cost is *totally useless* without some additional constraints so that you're measuring average cost of *similar projects*. If you took all PVRs with roughly equivalent feature sets and examined cost based on embedded OS, *then* you might have a useful study. The current one is totally useless other than for FUD use.

  16. Not just embedding Windows, but coding on it by MichaelCrawford · · Score: 4, Insightful
    I'm not sure most people understand that Windows is the most widely used host operating system for developing embedded code. I haven't read the article, but I imagine the study also discussed the issue of which workstation OS was best for writing your code on.

    One reason nowadays that most embedded developers host their development on Windows is that most embedded tools publishers only make their tools available to run on Windows. Often these tools, like compilers for arcane chips, are quite specialized, so the developer is left with no choice.

    That said, I think whoever wrote the report is on crack if they think Windows is a better development environment than Linux. I have been doing embedded for a year now, and one of the main things I still dislike about it is that I have to do most of my development on Windows.

    It is quite common for each compiler vendor to write their own integrated development environment, with an editor and integrated build system. But the market for these products are not as great as the market for IDEs for the development of desktop or server software, which means they can't invest in developing a more refined GUI for their IDEs, so their basic usability and quality is quite poor.

    If you think Visual Studio is a lousy environment for development, you should try the ARM IDE or TI's Code Composer Studio. Using them is like pounding nails with your fists.

    However, the situation is slowly starting to look up. GCC targets many embedded CPUs and is starting to become widely used for embedded development. The other GNU tools also form a more or less complete set of what you would need to develop embedded products, with GDB acting as both a debugger and simulator, LD able to function as an embedded linker, being able to do two-machine debugging with GDB and so on. Also there's GNU make, CVS and so on.

    The result is that while I had to use the proprietary (and expensive) ARM compiler to develop for the Oxford Semiconductor ARM7TDMI-based 911 FireWire/IDE bridge chip (which allows you to hook up inexpensive IDE disk drives as firewire storage), they switched over to building their firmware with GCC for the 922 USB/FireWire/IDE 922 bridge chip.

    I've been using GCC under Cygwin for my 922 development, but a CD with a new SDK on it is expected to arrive in the mail any day now. When it does, I will have a choice of Cygwin, Linux or Mac OS X development environments, all running the same version of GCC. And I'm very happy about that.

    Most likely, though, I will use Mac OS X for my 922 development. I'd prefer using Linux to Windows, but if I can use Mac OS X, I'd prefer that to Linux, if for no other reason than the fact that the clipboard works correctly, as well as that I could use CodeWarrior to edit my source.

    Maybe if I get real ambitious, I might write a CodeWarrior plugin so I can use the CodeWarrior IDE to compile my code with arm-elf-gcc.

    (And don't give me crap about not using Emacs. I was an expert at Emacs when most of you were still in diapers. I still have my .emacs file which I first created in 1987. But I vastly prefer CodeWarrior's GUI text editor unless I have some reason to run a bunch of Emacs lisp code on my source file.)

    --
    Request your free CD of my piano music.
  17. That was the point... by barawn · · Score: 4, Insightful
    The "I have experience that differs" is the point of the article. He's an embedded systems consulting company. It's his job to have experience. If it was in context (rather than simply point, click, read through Slashdot) it might make more sense.

    Check here for a more thorough "factual" rebuttal, including my favorite quote from the original report...
    For the purposes of runtime royalty comparison, only Windows CE .NET and embedded Linux will be considered.

    No reasoning, no nothing, as to why Windows XP Embedded (which a lot of the reasoning of the rest of the report was based on). Why, might one ask, would someone do this? Might it have something to do with the fact that the royalty cost for Linux is $0, the royalty cost for Windows CE (in volume) is $2.60, and the royalty cost for Windows XP Embedded is approximately $100 per system?

    Yah. OK. That's a bit like me saying I'm going to compare the reliability of Toyotas and Fords, but for the purpose of the study, only Toyota cars that don't actually run will be used.

    I mean, really - the original report is so bad it's laughable. It really didn't even NEED a rebuttal.
  18. Re:Hold it right there you scumbag! by bninja_penguin · · Score: 3, Informative

    I know software crashes occasionally, no matter who wrote it, but man, you ought to get your power checked or something, I mean holy shit man, My TV crash? NEVER! All three of my cell phones? NEVER! I don't have a DVD player, but my cable box? NEVER! My router? NEVER! I do turn it off when i leave for overnight or longer trips. WAP? I don't have one, but my cordless phone (landline?) NEVER! I don't have a Satellite Receiver either, but my microwave? NEVER! My automatic sprinkler system controller? NEVER! My coffee pot? NEVER! My VCR? NEVER! I could go on, but what's the point?

    Yeah, he should check out his hardware/driver setup on his 2003 box, but you also need to remember 2003 has just been released, and is, for all intents and purposes,still BETA. Until it's been run in the field for a year or so, and the bug reports have been submitted back to Microsoft and fixed, it is BETA. Just try getting Exchange Server up and running properly, and with all the functionality of it running on 2000 server. It's a bitch, and even Microsoft admits that.

    --
    For those who describe their systems as 'boxen', do you order multiple 'boxen' of corn flakes also?