Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
Power Ubuntu Windows Linux

Can Ubuntu Linux Consume Less Power Than Windows? 225

An anonymous reader writes "Now that the big Linux kernel power regression has been solved it looks like Ubuntu 11.04 can compete with Microsoft Windows 7 in terms of overall power usage. New tests revealed by Phoronix show the power consumption of Ubuntu 11.04 vs. Windows 7 operating systems. On a range of different systems, the power consumption of the Linux OS was comparable to that of Windows except for a few select workloads and systems."
This discussion has been archived. No new comments can be posted.

Can Ubuntu Linux Consume Less Power Than Windows?

Comments Filter:
  • Any more questions?
  • by Kagetsuki ( 1620613 ) on Wednesday June 29, 2011 @12:20PM (#36612304)

    It is possible it could consume less power, but that doesn't necessarily mean it always does. Different hardware, specialty drivers, default settings vs tweaked settings - come on?

    • From personal experience (my Toshiba laptop), I noticed that Ubuntu consumed more power than windows Vista (based on how long the battery would last). I tried to tweak both of them to run on more optimal settings for battery life. I probably did a poorer job tweaking the Ubuntu settings though due to my own lack of experience with the OS. This was running the same basic programs in both OSes (office programs like powerpoint/word and the openoffice equivalents, mozilla/firefox with lots of tabs open)--not pa
  • by spiffmastercow ( 1001386 ) on Wednesday June 29, 2011 @12:24PM (#36612344)
    Ever since I upgraded my netbook to Ubuntu 11.04, it crashes randomly and often. I'm talking more that Windows 95 with no patches. Hell, more than Windows 3.0. While solving power management would be nice, it's a moot point if the computer is always off because I can never use it.
    • Define crash? Random apps crashing, the UI crashing, or a full blow reboot required (assuming the system isn't doing it on its own)
    • by vbraga ( 228124 )

      I've had a few kernel panics (UI freezes and caps lock lights keeps blinking) since I upgraded to 11.04. It doesn't happens when I'm running Windows.

      • And while that shouldn't happen at all, at least with Ubuntu you can often recover from such a state by switching to a terminal and restarting Compiz and/or X. Not the case with Windows.

        • by amliebsch ( 724858 ) on Wednesday June 29, 2011 @01:38PM (#36613390) Journal

          Really? I never you could recover from a kernel panic by switching to a terminal and restarting X! Mod parent INFORMATIVE!!!++ Thank you so much!!

    • 11.04 seems to include a kernel with LOTS of regressions, or the Ubuntu maintainers added some to the kernel/modules packages.

      For example, the wireless drivers for Ralink RT2860 chipsets were rock solid from 9.04 to 10.10, but were completely broken after an 11.04 update. Even after doing some module blacklist magic, the wireless drivers now perform horrifically and fail to connect very often.

    • You know seriously, I used to have crashing problems like pretty much everyone else but two things have changed:

      1. My use of Windows is limited to "what I need." I don't install crap of any kind. If I don't use it, it doesn't get installed or will soon be removed if it was there.
      2. Windows uptimes have increased for me over the years. Perhaps it's all the bug fixes and what have you, but whatever the case, I don't have as many problems. (Other users, however, still seem to have the same problems, so it'

      • It's not all to MS's credit? I thought the OS was there to provide resources to applications. One example might be the "still on" resource.

    • by mcelrath ( 8027 )

      I've been having kernel panics regularly, I recently figured out that it's due to "hardware acceleration" in the binary flashplayer, coupled with the open source video drivers (ati, for me). Right click on a flash animation and turn off hardware acceleration. And/or un-install that steaming pile of dung. Still, it shouldn't be causing kernel panics. I think the open source/drm drivers need work.

    • As I recall windows 3(? maybe I'm thinking of 3.11, but I think I'm thinking of 3) was damn hard to crash...

    • That's interesting. My workstation at work is running Ubuntu 11.04 and hasn't had a single issue at all. I do Android development for a living and when I first got this job, they plunked me down in front of a shiny new Win7 box.

      I gave it the old college try for two months but finally, after gritting my teeth for the last time, I made a beeline for the blank CD's and had Ubuntu installed over my lunch break.

      Now, I can finally be productive with great tools integrated into the OS. SSH, bash, apt-get, tilda

    • by 21mhz ( 443080 )

      Might it be thermal shutdown? I had this issue after upgrading; apparently the new kernel does not play well with the (probably buggy) fan controller on some ThinkPads. After I blew the CPU fan slots through, it did not overheat any more.

  • Can Ubuntu Linux Consume Less Power Than Windows?

    Shouldn't it already? It's not like anything in Linux is causing the 3d acceleration to kick in.

    • What about the compositing window managers (like Compiz)?

      • Which will save power. The gfx card, as it happens, is rather good at doing window translation and compositing in hardware, seeing as that's what we use it for. A properly-designed gfx card (in a low-power state) and compositing WM should blow the pants off of a CPU solution. I don't know if this is the case (it depends on how far back an unloaded GFX card will throttle) but I wouldn't be super surprised

  • Both operating systems are written in C, no? If thats the case, they should be able to compete. I would assume MS devs have an advantage though, because they can evaluate opensourced code to see where it's efficient. They wouldnt have to rip code, but it gives an idea of what can be done better

    I'd also wager that efficiency isnt solely in the hands of the ms/nix devs. If someone writes a shitty firmware for a hdd or disc drive that doesnt take power consumption into account its hardly the OS dev's fault
    • Both operating systems are written in C, no? If thats the case, they should be able to compete.

      Quote of the day...

    • Both operating systems are written in C, no? If thats the case, they should be able to compete.

      Ford F-150 and Smart 4Two both run on gasoline, no? If that's the case, they should be able to compete on MPG.

      • Your analogy makes no sense at all.

        F-150 and 4two running on gasoline is akin to a PC running on electricity. A more apt comparison might have been F-150 and 4two both using an IC engine, and i dont think thats the case with 4two.
        • A more apt comparison might have been F-150 and 4two both using an IC engine, and i dont think thats the case with 4two.

          What kind of engine would you expect it to have that runs on gasoline, but isn't internal combustion?

          If you meant to say that it's a hybrid, then no, it's not (at least not the conventional variety).

  • I've run Debian derivatives going back to '06 on my laptops -- an HP, a Dell, and a Samsung (this was the point at which I could install Linux and not have to spend the next several hours getting the network card and wireless card to work with my existing hardware). However, I found that Linux consistently cut my laptop power by about 20-30% over Windows XP. Vista was worse, of course, as that had serious power issues on laptops at first, but now Windows 7 performs as good as or better than XP, as near as

    • by hitmark ( 640295 )

      I suspect this is because, as found by Phoronix, Linux is unable to turn hardware off when it is not in use. This thanks to buggy ACPI or similar that the OEMs work around in their own drivers for Windows, but that the Linux devs have to find out about the hard way. Hell, not too long ago there was a desktop motherboard that was unbootable if Linux was honest about itself. This thanks to a garbage ACPI entry for anything other then Windows.

      • Do you find the same trend to be true just for hard disks? I've got a couple of Acer laptops and my wife uses hers for hours every day. And every day, the hard drive consistently reads >50 degrees C. I'm concerned that the hard drive will have a shorter life because of this.

        Any suggestions as to what I can do? Would an SSD be better?

        • Um, that's probably a matter of poor cooling in the Acer laptop. There might be a lower-power mode you could set with hdparm, but spinning the disk down completely just isn't practical: it gets spun back up too much. And a good SSD might cost more than the laptop is worth--certainly more than a new hard disk.

          Buy a new hard disk if you want, one that's quiet and cool--but that's no guarantee it will be cool in your laptop.

          My suggestion: back up your data regularly, set aside some money to buy a new disk wh

  • Is it just me or did they not include Aero for the assessment?

    The video hardware is more efficient at rendering than the CPU, so this could skew the results quite a bit by potentially having Aero off.
    • This is likely even more of a concern on the combination GPU/CPU systems where Intel/AMD are banking on the fact that you'll be rendering entirely on the GPU.
      • It could be worse, but given that things seem to be going in the Open CL direction, Linux should have the same access that MS does to the tools necessary to make that work.

        • Well, I mean, I'm not saying that they shouldn't be equal, but the reality is they aren't. Tests like this don't tell the whole story because someone feels the "test isn't equal" if they don't.

          The regular person doesn't know that. This article hitting slashdot will reach a lot of people that won't know the difference. It should be a test of what you experience.
  • In my experience, it depends on the hardware to some extent.

    For example, consider that newer laptop GPU setups (using NVIDIA Optimus and whatever ATI calls their equivalent) use "switchable graphics." Essentially the output device is always a cheap integrated device, but when real GPU power is needed the OS will seamlessly switch over to a separate, bona fide GPU and have its framebuffers forwarded to the integrated chip.

    This requires kernel-level support for the switchable graphics systems -- suppor
  • Just don't give me any "base system" crap, test it with a real system running real background tasks. A windows machine running anti-virus real time protection may add significantly in terms of power. Multiplied over an entire office, that could add up. But I dunno what they used, cause I think the Phoronix website needs a little more power to withstand /.

    • Most AV suites are active less than 1% of the time on modern systems. Yes, they do add *some* cost, but it's hardly likely to sway the results by much.

      Besides, AV is not strictly required. If you know what you're doing and don't get particularly unlucky, you can get by for years without it. If you are sufficiently unlucky, you can get completely taken over even if you know what you're doing and run up-to-date AV as well. I run a free and non-obtrusive AV program, but if it started having a noticable impact

  • by gamrillen ( 1972402 ) on Wednesday June 29, 2011 @01:07PM (#36612976) Homepage
    Whenever Canonical releases a new version of Ubuntu, I'm always game to take it for a test drive. I use an EeePC 1005HA netbook and a VirtualBox VM (Windows 7 x64 host) to do my testing. For the last three versions (10.10, 11.04, and 11.10) I've had issues with the netbook overheating and shutting down the hardware. Additionally, the sleep/hibernation functionality never seems to work just right. Sometimes, when I close the netbook, it won't go to sleep at all and the LCD screen will stay turned on. Other times, the netbook will sleep peacefully, but won't boot back up when I open the lid (as set in my preferences) or hit the power button. I have to remove the battery and do a hard boot. As for the VM, Ubuntu runs incredibly slow even with the guest additions installed. I have to sometimes triple click on single click buttons to select something, and Gnome likes to generate random error messages. On the flip side, I can run Windows 7 x32, Windows 7 x64, and even Windows XP x32 on the netbook, and won't have any of the issues I see with Ubuntu. The same goes for using the three Windows variants mentioned above in the VM. Yes, less power consumption is a great thing, and yes it's awesome that interface tweaks are happening to make it prettier, but until stability issues with fairly common chip-sets are resolved, I won't be using Ubuntu on a daily basis. However, Linux Mint, which is based on the most current stable release of Ubuntu seems to take all of Ubuntu's shortcomings and clean them up. Mint just seems... tighter. Everything flows better, and I don't see the glitches that I normally see in Ubuntu.
    • by 0123456 ( 636235 )

      Weird. I have a 1005HAB running 10.10 and it's fine other than Unity randomly crashing when closing a Firefox window (which doesn't surprise me given all the other Unity bugs).

    • by Abreu ( 173023 )

      I have no problems with Ubuntu 11.04 suspend or hibernate on my Acer Aspire One (similar specs to an EeePC)

    • by rwa2 ( 4391 ) *

      Hmm, I was given the impression that stock Ubuntu is pretty dismal on netbooks. Try one of the netbook remixes that actually use the array kernel tuned for Atom chips and other netbook hardware.

      My favorite was eeebuntu [eeebuntu.org] 3.0 , but it hasn't been updated recently while waiting for the devs to polish off their new Aurora distro.

      I've also played with Fuduntu [fuduntu.org], which seems like a nice rpm-based distro, but my machine didn't survive a yum update. I might try again.

      Then there's the Ubuntu Netbook Remix, which prob

    • Um...

      Wouldn't sleep/hibernate functionality be the job of the hosting OS?

  • by RotateLeftByte ( 797477 ) on Wednesday June 29, 2011 @01:22PM (#36613172)

    There are far more usable and frankly higher quality distros than anything that comes from Canonical.

    I'll probably be shot down in flames but as a long term linux supporter (since slackware 1.1 on Floppies) I've seen it evolve beyond all recognition.
    At first Ubuntu was a breath of freah air. It took the debian dinosaur and shook it alive. Now, they are changing things and IMHO not taking the user base with them. I know of at least 10 former Ubuntu fans who have jumped ship since 10.10 came out. The quality is just not there any more. Far too much is crammed into each release with little thought for fixing the bugs.
    Non of their stuff seems complete. Or in agile dev speak, 'It is not done.'
    This is totally wrong and is only storing up a vast reservior of technial debt for the future.

    Let the flaming begin.
    Anon coz I have to work closely with Canonical in my day job.

    • Call it what you want but Ubuntu looks to be the de facto OS that represents the Linux desktop market. Even when I ask my friends who are not familiar with too many Linux distros , the first thing they can come up is "Ubuntoo". As such, like it or not, Ubuntu is the only OS that I can think of that has the resources and the organization structure to compete with Windows. However, with the recent changes following 10.10, it seems that they're beginning to lose the competition. That's bad news for every other
      • We'll see how long that lasts, considering how recent decisions appear to be more about chasing people away than making a better distro. I was using Ubuntu, but I won't be using it anymore, considering how much crap I got when I upgraded last time. I'm sorry, but if you're going to have a release be unstable, the user should be given a heads up that they're upgrading to a development release.

    • RotateLeftByte (797477) says:
      Anon coz I have to work closely with Canonical in my day job.

      Uh oh. You're toast.

      To be more on point, perhaps I'll be shot down in flames with you, but you're clearly correct. Ubuntu is far from the best ("highest quality") desktop distro, and it doesn't really make for an ideal server OS either, in light of the great Linux-based alternatives. I'm glad Canonical and Ubuntu exist because some of the projects they work on which set them apart may actually have broad appeal and be good for the "community" in the long run, but it is too bad it's them who have become

    • Actually, I want to know more. I've been shopping distros and so far, have been working with Fedora for my laptops. Which one do you like so far?

      No flames here. Only curiosity.
    • by marnues ( 906739 )
      I don't think you use Ubuntu for the reasons that most people use Ubuntu. It is the user friendly version of Linux. I use it because I no longer care to do anything past install to correctly setup my machine. Ubuntu is the only flavor that has a fully functional setup for those of us that want to use the computer rather than playing sysadmin for a single user machine.
  • All hardware has VPD strings (vital product data) strings that will tell you what you have. Why cant they come up with a simple cheat sheet of default parameters for the hardware. Write some code that walks the bus from CPU out to the last usb dongle and if any gotcha hardware comes up, then its "magic numbers" get added to the boot parameters.
    If the hardware is trustworthy, fine...no parameters. If not, then it gets the fixes it needs. Are these rules being hard-coded in now? Why edit a file in a

    • See, that would be great, except it doesn't work quite like that: Generally such hardware configuration would be entered into the driver, which works great... Until you end up with two or more pieces sharing the same VPD, but need slightly different handling rules. Or, what happens if a piece of hardware just plain lies about it's capibilities? That's the real problem, and it's hard to work around: Some driver may work perfectly well on the dev's device, but on someone's newer/older revision(etc.) it doesn'

  • Now it's time for the people insisting that there was no power regression, to admit they were wrong, and their attitude would have kept this bug swept under the rug if they had their way. I saw a few people trolling Phoronix for the past month saying this regression didn't even exist.
    • Of course, it *might* have not happened on their machines! AFAIK, it was due to some faulty bios tables, and only from some MFGRs. So, one same-spec'd model might not have the problem, and others might.

  • by Billly Gates ( 198444 ) on Wednesday June 29, 2011 @03:32PM (#36614484) Journal

    Linux has seriously regressed over the past decade. I remember when Linux was a lite alternative to Windows 2000 and had power management in software that was orders of magnitude better.

    Those days are long gone.

    I am sick and tired of playing with releases of Ubuntu and Fedora hoping this one will truly be unique and beat Windows. Last March I switched back to Windows 7 as I do not have time to tinker and fool around with older releases of gnome to avoid unity/gnome-shell, and trying to enable hardware GPU accelerated web browsing experience. If you want it to perform as good as Windows, then just use Windows.

    • by hitmark ( 640295 )

      The reason for this is that there have been a transition from APM to ACPI. The former leaves the details up to the OS, the latter have some kind of meta-language for "describing" the hardware capabilities. Sadly it was set up while MS was still doing their EEE thing, and so we have something that on paper should be a standard, but in reality only really work in Windows. The rest of the world have to second guess everything.

      And EFI seems to be heading in the same direction...

  • The Ubuntu system stays on.

BLISS is ignorance.