Thursday, September 29, 2011

The truth about Linux Power Management "issues"

Anybody who's even remotely interested in Linux probably heard about a "power regression bug" in the Linux Kernel that was making lots of noise lately. The whole thing started from several posts at Phoronix, which not only stated the problem, but also accused the Linux Kernet team of completely ignoring it and doing nothing to fix it.

Power management is anything but a simple subject, so a big majority of us users hardly know enough to challenge claims about a power management bug. To make matters worse, many Linux users dual boot with Windows, which usually does a better job at energy saving (thanks to optimized proprietary drivers). For most of those users who could notice how their Linux installation ate their laptop battery faster than Windows, it didn't take much to give this rumor solid credibility. Therefore, the fact that Linux Kernel developers were apparently doing nothing about it simply felt all the more annoying, which only helped spread the rumor like fire on a windy day.

Well, guess what? That's all it is, a rumor. FEWT, the main developer behind the wonderful FUDUNTU distro and the slick Jupiter power management application has put together a very informative ARTICLE on this, clarifying some of the misconceptions spread by the whole power bug fiasco originator, Michael Larabel.

I very much recommend reading FEWT's article to understand this matter a bit better, and most importantly, to avoid spreading this poisoning rumor any further.

Thanks FEWT for putting the time to clarify this one!

9 comments:

  1. Sorry to disagree, but NO. It is a very real issue, on some laptops (using sandy bridge) the difference is 78%, so the battery only lasts 22% compared to windows. That is not a rumor.

    In my particular case, even before the phoronix guys writing the article I realized the life of my battery went from 2 hours to 50 minutes. Not to mention it was running -really hot-. This kind of posts saying "its a fiasco and you who are getting a portable instead of a laptop are crazy" are just eluding reality, sorry to say that.

    ReplyDelete
  2. @Luis Sanchez: I think you are misunderstanding my post and Fewt's. Nobody is claiming that Linux power management is at the same level as Windows, that would not be accurate. Linux lacks the great support Windows gets from hardware manufacturers which provide optimized drivers, plus the Linux Kernel has more than just mobile devices to keep in mind.

    What this post claims is that it is not true that power management has got worse because of that infamous "regression bug". First, because it is not a bug, and second because the battery life loss is marginal if it exists at all. What you describe there is a behavior that has unfortunately been there in Linux for a long time.

    NOTE: 78% difference is NOT normal, I recommend you check what you are doing because that's nowhere near the expected boundaries. I get 20% less on Linux, perhaps a smaller difference if I take the time to tweak my installation (ie, in KDE I would disable Nepomuk, Strigi, Akonadi, and any other daemon I don't need running, plus I would carefully set my battery saving settings just like Windows does, etc)

    ReplyDelete
  3. Absolutely not true, Jose Maria, whether it is Ubuntu, Debian, Suse, Slackware, whatever else, Linux, very luckily, can grant me at most 2.5 hours of work in my HP 8460p, while Windows can easily grant me from 6 to 7.5 hours, depending on the power scheme used. Laptop wise, I have learned that the best way to make the battery last under Linux is to leave the laptop off. Linux is great for desktop use or for laptops that are connected to the AC. Until Linux can match Windows' power performance, I shall use Linux in my laptop only as a curiosity. By the way, I understand that proprietary drivers make a huge difference, but I grant such an argument very little importance. Facts are what matters and the bare fact is that Linux is not match to Windows in laptops powered by their batteries.

    ReplyDelete
    Replies
    1. And what exactly does that have to do with my article? I will just reuse my answer to another comment above, because it mostly covers the same thing:

      "I think you are misunderstanding my post and Fewt's. Nobody is claiming that Linux power management is at the same level as Windows, that would not be accurate. Linux lacks the great support Windows gets from hardware manufacturers which provide optimized drivers, plus the Linux Kernel has more than just mobile devices to keep in mind.

      What this post claims is that it is not true that power management has got worse because of that infamous "regression bug". First, because it is not a bug, and second because the battery life loss is marginal if it exists at all. What you describe there is a behavior that has unfortunately been there in Linux for a long time."

      As for you not giving importance to the whole driver thing is, well, funny. If you compare power management features and there is a clear disadvantage on one side, it must be acknowledged, simply because that simply invalidates any of your claims. You see, Linux power performance is probably better than Windows´, it´s the support of proprietary devices that is worse. If you said so, you´d be closer to being accurate. Install Windows on a new box without installing any of the third party drivers/tools and then come back and compare it to Linux, that would be a meaningful comparison and it would show what the OS has to do in that scenario. You are comparing 3rd party power saving features (that are not Windows´) to Linux, it´s comparing apples to oranges.

      Last but not least, I think your computer has issues or you are not doing a good job at setting it up. I get almost 6 hours on my HP laptop with Kubuntu with all power saving features OFF (that is screen on at full brightness all the time), and around 10 hours with the default KDE power saving features enabled. As of today, with Kubuntu 12.04 and Windows XP, I find the difference to be down to a few minutes, so you really should double check what you are doing or which distro you are using, because your results are far from the current reality.

      Delete
    2. [quote]Last but not least, I think your computer has issues or you are not doing a good job at setting it up. I get almost 6 hours on my HP laptop with Kubuntu with all power saving features OFF (that is screen on at full brightness all the time), and around 10 hours with the default KDE power saving features enabled. As of today, with Kubuntu 12.04 and Windows XP, I find the difference to be down to a few minutes, so you really should double check what you are doing or which distro you are using, because your results are far from the current reality.[/quote]

      This is the same story all over the internet, support sites of linux distro's saying people are doing things wrong (its a clean install!). The only thing that is wrong in my opinion is tat certain hardware is not supported by the open drivers, and by the hardware vendor drivers. Leaving all laptops with some special setup without a good driver (controlling fans, gpu going 100% load all the time ect...). I have try ed hours and hours reading blog posts, and following so called fixes... Nothing has helped me. I've come to the conclusion that if u want to run linux on a laptop: buy one from the supported laptop list. Dont try it with your Acer(or whatever manufacturer) laptop made for windows (especialy those old (6yrs) or new Ati/Intel hybrid gpu systems.

      Delete
  4. I have used both Windows and Linux on my laptops, apart from the usual slowdown using Windows 7, the battery life has been a difference of 15% at the most give or take a few, if I use Jupiter, that gap goes down further to 5%, I can well live with that considering the multiple benefits I get with Linux. Make sure to use distros with latest kernels as tweaks are added regularly, I suggest semi rolling releases like Chakra, Sabayon etc and not Ubuntu LTS where kernel remains old, in that case use Ubuntu regular six month release. Newer kernels invariable make latops run better in my opinion due to various improvements and optimizations done.

    ReplyDelete
  5. I understand your post, it is not a bug, It is just because windows has better drivers. I have with my laptop 2.5 hours on windows. Been trying x10 sorts of distro's of linux, get me 25-30 mins of battery power. GPU getting 90°. Its a real pitty, really like the linux initiative, prob won't try it again the next 5+ years.

    ReplyDelete
  6. Drivers? Yeah, right. What do you think is the main power consumer in a computer? The keyboard controller? The hard drive controller? The USB controller? Sorry, but almost all of the power goes into the CPU and the display hardware.

    ReplyDelete
    Replies
    1. Euh?? And those components don't need drivers??

      Delete