Announcement

Collapse
No announcement yet.

Is Intel's PowerTOP Utility Still Beneficial In 2016 On Ubuntu 16.04 To Save Power?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Originally posted by uid313 View Post
    You get better battery, but maybe you get lower performance.
    So you probably want to do this on your laptop, but not on your desktop.

    I wonder if the Dell Sputnik "Developer Edition" laptop preloaded with Ubuntu is configured this way.

    but here you do not have to go to a special compromise = you have separate modes [and it is usually a few] for AC and battery Here is an example of work the mechanism [my authorship] APM 5.1 - power and performance management for Ubuntu/Kubuntu/... - on the following configuration:

    Comment


    • #12
      Originally posted by uid313 View Post
      You get better battery, but maybe you get lower performance.
      So you probably want to do this on your laptop, but not on your desktop.

      I wonder if the Dell Sputnik "Developer Edition" laptop preloaded with Ubuntu is configured this way.
      Yeah, using powertop's recommended settings on a workstation that needs to perform well can hurt. Which is why things like tlp have automatic switching between profiles when on a/c versus battery.

      And yes, I had to put in an exception for my keyboard to keep it usable, and I also ended up cranking the dirty write back value beyond the powertop recommendation as well (60s instead of 15).

      Comment


      • #13
        @ext73

        This is always board specific. One of my test boxes has got problems when i enable ASPM in the firmware. You can not say do this or that in general. Powertop can help you to configure whats working of course, but it is impossible to get a generic solution with lots of items enabled by default. For laptops you certainly want to maximize the battery life, but don't forget to disable a 2nd gfx chip while doing so...

        Comment


        • #14
          You have to test battery run time, not wattage usage. You can really lower the watts used on your computer by simply restricting the cores from running at their highest frequency. But you'll also spend more time executing code that way, so you won't necessarily get better battery lfie, depending on the task.

          Comment


          • #15
            Originally posted by Kano View Post
            @ext73

            This is always board specific. One of my test boxes has got problems when i enable ASPM in the firmware. You can not say do this or that in general. Powertop can help you to configure whats working of course, but it is impossible to get a generic solution with lots of items enabled by default. For laptops you certainly want to maximize the battery life, but don't forget to disable a 2nd gfx chip while doing so...
            I ensures You that is possible to globalize it. In the end, all over the world our www.netext73.pl solutions are already using 15 334 machines As for the graphics ... Nvidia is always turned off when not needed ... and the powertop shows that gets 100% power ? ... This "order" goes to all devices on my APM while working on AC - in this mode, the machine has to give everything ... unless, that instead of intel-performance mode is selected intel-powersave [for Intel i3, i5, i7 ] - for systems using cpufreq are 4 different modes of operation ... but that does not mean it has to use it - specifically here nvidia was ordered to = turn it off - I take care of the smallest details See:

            https://dl.dropboxusercontent.com/u/...ny/optimus.png
            Last edited by ext73; 15 March 2016, 04:21 PM.

            Comment


            • #16
              Originally posted by molecule-eye View Post
              You have to test battery run time, not wattage usage. You can really lower the watts used on your computer by simply restricting the cores from running at their highest frequency. But you'll also spend more time executing code that way, so you won't necessarily get better battery lfie, depending on the task.
              Was that meant to say, that it is the overal power use that matters, that is a slightly lower watts would be cancelled out by slightly longer run time?

              One of the tests was the critical power consumption at idle, so that's moot; in second it looks like the tests complete in the same time from power consumption line. Avoiding unwanted wake-ups ought improve a benchmark performance, because the CPU is focussed on doing the work, rather than servicing distractions.

              The battery life, adds variables and takes too long to practically test, but it makes sense if power top, helps reduce the amount of work done that battery life will be longer so reducing Watts is a useful measurement.

              Comment


              • #17
                @ext73

                If you want that I look at a website then it needs to be in English or German language. Btw. the default settings of a kernel need to run on most systems, this is certainly not optimal for power saving. But if the system crashes or has choppy USB input then you most likely wont use it.

                Comment


                • #18
                  Originally posted by molecule-eye View Post
                  You have to test battery run time, not wattage usage. You can really lower the watts used on your computer by simply restricting the cores from running at their highest frequency. But you'll also spend more time executing code that way, so you won't necessarily get better battery lfie, depending on the task.
                  Agreed, if CPU and RAM power consumption are completely linear against clock speed, you want to let the CPU run at max speed to get a job finished so that the whole system will return to idle/sleep.

                  Comment

                  Working...
                  X