Announcement

Collapse
No announcement yet.

Linux 3.6 To Linux 3.13 Kernel Power Consumption Tests

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

  • Linux 3.6 To Linux 3.13 Kernel Power Consumption Tests

    Phoronix: Linux 3.6 To Linux 3.13 Kernel Power Consumption Tests

    At the request of many Phoronix readers, here are some new battery power usage benchmarks on every recent Linux kernel release from Linux 3.7.0 to Linux 3.13 Git. Has an Intel "Ivy Bridge" Ultrabook's power consumption changed much due to the continuous kernel churn? Here's the answer...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    No test of idle power usage?

    #phoroniFAIL

    Comment


    • #3
      My HP Pavilion m6 1009tx with Intel i7-3612QM, 8GB RAM, Radeon 7670, 750GB 5400RPM disc has FINALLY become useful under linux since 3.11. The WiFi driver had the transfer speeds at kbits/s! Now it's KB =) Running on Mint 16RC, up from Mint 15. Using only PPA's from Ubuntu's mainline kernel (3.12 kernel) and oibaf's for the mesa stack and Intel driver.

      I still cant yet properly suspend or hibernate, but that's likely an issue I can fix myself if it's not actually my fault (misconfig somewhere....)
      Battery life is pretty average, too (8 hours) the Catalyst drivers to access the Radeon. Obviously I'd want to use the Intel (along with Oibafness) driver for 'work' tasks, especially now the stack's matured well. It just sucks having to reboot compared to the 6 hours I was getting on W8. 3-4 with the Radeon =P Still, the advancement has been pretty good.

      Well, there's my two centsies on my recent experience =D
      Hi

      Comment


      • #4
        Originally posted by smitty3268 View Post
        No test of idle power usage?
        The testing is from idle state and idling is between benchmarks so you can look at the data between there.
        Michael Larabel
        https://www.michaellarabel.com/

        Comment


        • #5
          "OS X still does a superior job on delivering the longest battery longevity"

          Is there a measurable difference? And where does the extra consumption come from? I wonder whether it is mainly down to drivers not taking advantage of deeper sleep states or whether there are some applications on Linux which wake the processor more often?

          Comment


          • #6
            Originally posted by chrisb View Post
            "OS X still does a superior job on delivering the longest battery longevity"

            Is there a measurable difference? And where does the extra consumption come from? I wonder whether it is mainly down to drivers not taking advantage of deeper sleep states or whether there are some applications on Linux which wake the processor more often?
            Each macbook comes with a piece of the soul of Steve Jobs. Didn't you know?

            Comment


            • #7
              Originally posted by Michael View Post
              The testing is from idle state and idling is between benchmarks so you can look at the data between there.
              So it looks like 3.10 does have a significant power regression? It's hard to tell whether the poor scores there are real or just random noise since you didn't extend the idle tests.

              Comment


              • #8
                "OS X still does a superior job on delivering the longest battery longevity"

                Comment


                • #9
                  It would be nice if you tested longer idle periods. Also, comparing with Windows 7/8 would be good too. My Ivy Bridge notebook idles around 6 W - 7 W under Xubuntu 12.04 using the 3.8 kernel while under Windows 7 it idles around 5 W - 6 W. My notebook spends most of its time idling while I'm using it so I think there should be a little more focus on the subject. Windows gets generally lower average power consumption under various loads as well.

                  I hope you'll look into Linux vs Windows 8 power consumption more. Like with GPU performance, there's a big gap between power consumption under Linux and Windows. Too bad it can't really be isolated to one thing like when testing GPUs. Also, if you ever get a Haswell based notebook and it supports the new S0ix state then maybe see if the chip ever enters this state. When idle, it supposedly can provide suspend levels of power consumption without actually having to suspend the notebook.

                  Comment


                  • #10
                    Originally posted by Michael View Post
                    The testing is from idle state and idling is between benchmarks so you can look at the data between there.
                    You should still do a proper test and give some figures for average power usage during light work (e.g. text editing). Reading them off your graphs is a chore.

                    Comment

                    Working...
                    X