Page 1 of 2 12 LastLast
Results 1 to 10 of 18

Thread: Linux Kernel Power Management Targeting Memory

  1. #1
    Join Date
    Jan 2007
    Posts
    14,590

    Default Linux Kernel Power Management Targeting Memory

    Phoronix: Linux Kernel Power Management Targeting Memory

    One of the areas of hardware power management that can yield a surprising amount of power-savings but is often overlooked comes down to the system memory. Fortunately, new Linux kernel patches continue to be written for improving the Linux kernel RAM power management...

    http://www.phoronix.com/vr.php?view=MTM0NzU

  2. #2
    Join Date
    Dec 2011
    Posts
    2,023

    Default

    ARM draws little power, so 6% power save show up as significant.
    However, on ARM systems such as smartphones and tablets, what draw the most power is the screen.

    On x86 processors, the power saved would be negligible.

  3. #3
    Join Date
    Feb 2008
    Location
    Linuxland
    Posts
    5,049

    Default

    Say that to the caching server that has no screen and 1TB of RAM

  4. #4
    Join Date
    Jun 2009
    Posts
    1,119

    Default

    Quote Originally Posted by uid313 View Post
    ARM draws little power, so 6% power save show up as significant.
    However, on ARM systems such as smartphones and tablets, what draw the most power is the screen.

    On x86 processors, the power saved would be negligible.
    in a regular x86 Desktop would be negligible but it could help to extend RAM life, now on ARM and x86 Servers gain could be very important especially in high density setups

  5. #5
    Join Date
    Dec 2010
    Location
    MA, USA
    Posts
    1,263

    Default

    Quote Originally Posted by uid313 View Post
    ARM draws little power, so 6% power save show up as significant.
    However, on ARM systems such as smartphones and tablets, what draw the most power is the screen.

    On x86 processors, the power saved would be negligible.
    I think it depends on how this works. For example, I'm pretty sure you can't (effectively) tell a particular column in a single memory chip to operate in a lower-power state while the rest of the IC is at full power. However, you might be able to lower the power consumption of an entire memory chip in a DIMM, but perhaps the entirety of DIMMs are what get affected. Anyways, nearly all ARM SoCs involve a single IC for RAM, and if I'm right about individual columns/rows being unable to change their power state, then ARM systems will get a 0% benefit from this.

    I feel like the people who would benefit from this most are those who have 4+ DIMMs and are either overclockers or use ECC memory. Considering the overkill amount of RAM people buy these days, entire RAM modules could be completely ignored by the OS, which would save a decent amount of power.
    Last edited by schmidtbag; 04-10-2013 at 10:20 AM.

  6. #6
    Join Date
    Jun 2011
    Posts
    876

    Default

    Quote Originally Posted by schmidtbag View Post
    Considering the overkill amount of RAM people buy these days, entire RAM modules could be completely ignored by the OS, which would save a decent amount of power.
    But.... RAMDrive....

  7. #7
    Join Date
    Sep 2008
    Location
    Vilnius, Lithuania
    Posts
    2,538

    Default

    Quote Originally Posted by Luke_Wolf View Post
    But.... RAMDrive....
    It's called an SSD

  8. #8
    Join Date
    Jan 2009
    Posts
    1,345

    Default

    Quote Originally Posted by uid313 View Post
    However, on ARM systems such as smartphones and tablets, what draw the most power is the screen.
    While that is true, when you look at good reviews of devices, you see significant differences in phones/tablets with similar screen sizes.
    Apple sports a pretty impressive power management system (across all their devices) that shows up even when correcting for screen power draw. So, looking at other parts of the system can make a huge difference (more than this 6%, even).
    Look at the latest big Anandtech review of the HTC One. The iphone 5 tops, or near the top, for every battery test they have the least amount of control over (telephony being a part they just have to accept from whoever makes their telephony chips).

  9. #9
    Join Date
    Feb 2008
    Location
    Linuxland
    Posts
    5,049

    Default

    Quote Originally Posted by GreatEmerald View Post
    It's called an SSD
    Screw your SSD, my ram drive is many times faster than the SATA-3 bus

  10. #10
    Join Date
    Dec 2010
    Location
    MA, USA
    Posts
    1,263

    Default

    Quote Originally Posted by Luke_Wolf View Post
    But.... RAMDrive....
    RAM drives are nice, I even created an automated script to generate them, but I have yet to find a real practical use for them aside from live CDs. If there was a way to attach the RAM drive to at least 1 SATA port (but preferably more so I can do RAID 0) then RAM drives would be fantastic because then I could store an OS on them and boot from another computer. Imagine that though - a 4x RAID 0 RAM drive on SATA 3 - that's going to offer some performance you just simply can't beat.

    I'm really surprised someone hasn't created a SATA "bridge" like this yet. Or, a RAM drive that works this way. I'm sure it'd be wildly popular. The only true RAM drives that were made were SATAII and limited to like... 8GB. Modern SSDs are better, and cheaper.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •