Announcement

Collapse
No announcement yet.

ACPI, Power Management Get Big Linux 3.13 Updates

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

  • ACPI, Power Management Get Big Linux 3.13 Updates

    Phoronix: ACPI, Power Management Get Big Linux 3.13 Updates

    The Linux 3.13 kernel merge window has become even more exciting even though it's not even officially open yet. Besides many to-be-merged changes talked about on Phoronix already, the ACPI and power management pull is particularly exciting for ARM and Intel Linux users...

    http://www.phoronix.com/vr.php?view=MTUwODA

  • #2
    APUs need more love

    I'd really like to see better power management for APUs. I have a little HP DM1Z running an E2-1800 processor and it runs slow and hot even under XFCE which is suppose to be lighter on ressources. Meanwhile it runs really well in Windows which tells me it's not some dusty fan problem causing throttling. Granted my distro isn't on the latest kernel, but currently using 3.8 but still. A lot of improvements are still needed to these types of processors which are rapidly becoming the dominant processor for mainstream computing.

    Comment


    • #3
      use kernel 3.12

      Originally posted by mithion View Post
      I'd really like to see better power management for APUs. I have a little HP DM1Z running an E2-1800 processor and it runs slow and hot even under XFCE which is suppose to be lighter on ressources. Meanwhile it runs really well in Windows which tells me it's not some dusty fan problem causing throttling. Granted my distro isn't on the latest kernel, but currently using 3.8 but still. A lot of improvements are still needed to these types of processors which are rapidly becoming the dominant processor for mainstream computing.
      use kernel 3.12 with dpm

      Comment


      • #4
        Originally posted by mithion View Post
        I'd really like to see better power management for APUs. I have a little HP DM1Z running an E2-1800 processor and it runs slow and hot even under XFCE which is suppose to be lighter on ressources. Meanwhile it runs really well in Windows which tells me it's not some dusty fan problem causing throttling. Granted my distro isn't on the latest kernel, but currently using 3.8 but still. A lot of improvements are still needed to these types of processors which are rapidly becoming the dominant processor for mainstream computing.
        You need at least kernel 3.11 plus radeon.dpm=1 boot option for dynamic power management.

        Comment


        • #5
          Hey, what could be worse than a two minute edit limit ?

          A ONE MINUTE EDIT LIMIT. AUGGH !!!!!

          Comment


          • #6
            I recently discovered the timer slack controller for timers and really don't understand why this isn't made use of. Systemd has an interface for it (you can change timer slack in /etc/systemd/system.conf), logind is made aware of system state changes but there's no way that I've found to dynamically change the timer slack based on system state. By default timer slack is left unset.
            The reason I bring this up is b/c timer slack is supposed to have the effect of allowing timer-based events to wakeup less precisely which enables the scheduler to coalesce the events into fewer wakeup period thus reducing the number of times the cpu would have to exit its low power states.

            Comment


            • #7
              Originally posted by bridgman View Post
              Hey, what could be worse than a two minute edit limit ?

              A ONE MINUTE EDIT LIMIT. AUGGH !!!!!
              I feel your pain Bridgman -_- Michael i like the new font and texting styling, it looks a lot nicer. But the one minute edit delay is a big No-No

              Comment


              • #8
                For the ULV Haswell and Bay Trail chips, does the Intel P-State driver also tell the CPU to go into the new S0ix active idle state when idle? This new state is being touted as the main reason why ULV Haswell, Clover Trail and Bay Trail based Windows 8 devices get noticeably longer battery life compared to older ULV chips.

                Comment


                • #9
                  pstate :/

                  Originally posted by guido12 View Post
                  For the ULV Haswell and Bay Trail chips, does the Intel P-State driver also tell the CPU to go into the new S0ix active idle state when idle? This new state is being touted as the main reason why ULV Haswell, Clover Trail and Bay Trail based Windows 8 devices get noticeably longer battery life compared to older ULV chips.
                  i have a ulv ivy and p-state is really bad for battery life i don t know about haswell, i never see my cpu in 800 mhz (minimal) even in idle, because of this i disable p.state

                  Comment


                  • #10
                    Originally posted by Andrecorreia View Post
                    i have a ulv ivy and p-state is really bad for battery life i don t know about haswell, i never see my cpu in 800 mhz (minimal) even in idle, because of this i disable p.state
                    When P-States are enabled, the CPU-frequency does correlate to power use directly (i.e. you should not pay attention to CPU frequency to determine power usage). To tell whether P-States are working correctly use powertop and check that the cores are in C7 state most of the time.

                    Comment

                    Working...
                    X