Announcement

Collapse
No announcement yet.

Linux 3.13 Kernel Power Consumption Benchmarks

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

  • Linux 3.13 Kernel Power Consumption Benchmarks

    Phoronix: Linux 3.13 Kernel Power Consumption Benchmarks

    With word recently of a 50 Watt Linux kernel power regression that's still being investigated, I carried out some more power consumption tests of a Core i7 4770K Haswell system to see if its power usage has been impacted by recent kernel upgrades...

    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
    Ubuntu 13.10 and 14.04 versus Windows 8 power consumption in various conditions would be very intresting I think. Did the gap closed from 2012?

    Comment


    • #3
      Was intel_pstate enabled? For me it caused my laptop produce more heat than with acpi (From 46? to 53?), and thus more power consumption.

      Right now my linux setup uses less power than windows 7, by the way.

      Comment


      • #4
        Got a Lenovo t440p Laptop with a Haswell i7-4600M and the System comsumes twice as much power (16 Watts using Intel P-State driver) during idle state than my old T420 with a I7-2620M both without discrete graphics and all Powersave options enabled and 3.13-rc4 kernel. Very disappointing results for a CPU that's on the market for more than 7 months! I hope that Michael will be able to run some Power Consumption tests on mobile CUP's.

        Comment


        • #5
          If/when you get Haswell Core i U/Y series and Linux compatible Bay Trail notebooks, I'd like to see comparisons with Windows 8 in terms of idle power consumption and performance per watt numbers. I'd like to see if the claims of Windows 8 making "full use" of Intel's S0ix active idle state and low latency selective component power down/up features (currently only works on Connected Standby capable Atom devices) really make a big difference.

          Comment


          • #6
            Originally posted by guido12 View Post
            If/when you get Haswell Core i U/Y series and Linux compatible Bay Trail notebooks, I'd like to see comparisons with Windows 8 in terms of idle power consumption and performance per watt numbers. I'd like to see if the claims of Windows 8 making "full use" of Intel's S0ix active idle state and low latency selective component power down/up features (currently only works on Connected Standby capable Atom devices) really make a big difference.
            I second that!

            Comment


            • #7
              Originally posted by alazar View Post
              Was intel_pstate enabled? For me it caused my laptop produce more heat than with acpi (From 46? to 53?), and thus more power consumption.

              Right now my linux setup uses less power than windows 7, by the way.
              Same issue here on both my IVB and Haswell laptops, heat goes up with pstate enabled, also peak temps are higher unless I use Thermald. With ondemand temps are lowered and I also don't find any noticeable power savings with pstate. With both CPUs, the battery consumption matches Windows 8.

              Comment


              • #8
                Originally posted by guido12 View Post
                If/when you get Haswell Core i U/Y series and Linux compatible Bay Trail notebooks, I'd like to see comparisons with Windows 8 in terms of idle power consumption and performance per watt numbers. I'd like to see if the claims of Windows 8 making "full use" of Intel's S0ix active idle state and low latency selective component power down/up features (currently only works on Connected Standby capable Atom devices) really make a big difference.
                Just registered to say that I'd like to see this too on a Bay Trail device once Intel/MS get Connected Standby working on 64 bit Bay Trail. The claimed tight component requirements for Connected Standby certified devices supposedly provides very long battery life under Windows 8. I'd like to see if these Connected Standby certified platform claims are true and what differences they make compared to Linux.

                Comment


                • #9
                  Originally posted by riqtik View Post
                  Just registered to say that I'd like to see this too on a Bay Trail device once Intel/MS get Connected Standby working on 64 bit Bay Trail. The claimed tight component requirements for Connected Standby certified devices supposedly provides very long battery life under Windows 8. I'd like to see if these Connected Standby certified platform claims are true and what differences they make compared to Linux.
                  Ya, I keep hearing that making something Connected Standby capable has very strict hardware requirements which lets Windows 8 have very fine control over each component and power consumption without affecting overall performance much. Too bad I haven't seen anyone verify any of this since you can't really toggle whatever Windows 8 does on/off but if it's true then I'd like to get these benefits under Linux as well on the same device if they're not already available.

                  Comment

                  Working...
                  X