What should "sleepstudy" show for a Surface Pro 3

Anonymous
2014-07-02T16:59:58+00:00

I'm interested to know what other users tend to see when running the "powercfg /sleepstudy" tool on their Surface Pro 3s (as per http://blogs.windows.com/windows/b/windowsexperience/archive/2014/06/26/sleep-study-diagnose-what-s-draining-your-battery-while-the-system-sleeps.aspx).

I'm finding that all my "InstantGo" sessions have a drain rate of between 1200 and 2200mW, which can correspond to between 10% and 20% of the battery drain in the 4 hours before the machine hibernates.  

Looking at the culprits, almost all the sessions show the following device running for the entire "sleep" period:

USB xHCI Compliant Host Controller (\_SB.PCI0.XHC) Fx Device 100% 4:00:00

Is this expected?  FWIW, I tend to leave the keyboard cover connected during sleep.

Sometimes (when the drain is at the high end of the range), I also see the following:

Intel(R) HD Graphics Family (\_SB.PCI0.GFX0) Fx Device 100% 4:00:00

Interestingly, during the first few days I had the SP3, the graphics driver was appearing in every session of the sleep study report.  More recently, though, it has stopped appearing.  Recently I've been leaving the Surface at the start screen rather than the Desktop, so this may be related - is there a recommendation about coming out of the Desktop before going to "sleep"

I also never see anything in the "DRIPS histogram" - should I?

What are other people seeing?

On a related note, does anyone know if there is a way of changing the time before the machine hibernates?  For me, 2 hours (or even 1 hour) would probably be fine - if I'm actively going from one office to another, then the "instant" resume is great, but if I'm at home, leaving the machine for a while, then 4 hours is overkill.

Surface | Surface Pro | Power and battery

Locked Question. This question was migrated from the Microsoft Support Community. You can vote on whether it's helpful, but you can't add comments or replies or follow the question.

0 comments

82 answers

Sort by: Most helpful
Most helpful Newest Oldest
  1. Anonymous
    2014-10-08T15:12:40+00:00

    :) Hopefully it gets fixed soon!

    Was this answer helpful?

    0 comments
  2. Anonymous
    2014-10-08T14:52:56+00:00

    That is a fair point and I am jealous!  I do see something nearer the 400mW on the SP3 I have at home, but with the work one I started off losing about 20% over 4 hours, so getting that to 10% is definitely a step in the right direction!

    Was this answer helpful?

    0 comments
  3. Anonymous
    2014-10-08T13:49:21+00:00

    I'm glad it is improving for you. But 904mW discharge rate is still very high. My discharge rate when everything is green is about 136mW, and I lose barely 1% of battery during 4 hour sleep before it hibernates.

    Was this answer helpful?

    0 comments
  4. Anonymous
    2014-10-08T08:44:24+00:00

    I think we're definitely on to something here.

    I closed down both Skype and my company VoIP phone client last night (both of which have access to USB headset and camera) and, lo and behold, for almost the first time ever I had sleep study show "all green" in terms of activity (no xHCI hub and just a few things active for around 5% of the time).  I also had a large bar in the 32s part of the DRIPS graph (which is normally completely blank).

    What is also interesting is that during this session, I didn't actually lose that much less power than in the previous session where the xHCI device was active - I had a discharge rate of 940mW compared to 1031mW the previous day.  I think this tends to tell me that (in my case)

    -  closing these comms devices does help

    -  my device has actually settled down since I started this thread so that even with the xHCI active, I don't lose more than around 10% per 4 hour standby period.

    Was this answer helpful?

    0 comments
  5. Anonymous
    2014-10-07T18:42:45+00:00

    Adding the usage chart that shows pre/post Skype/clearsea removal (removal stats starting at row 6)

    1 2014-10-04<br>09:20:49 2:00:05 2,202 mWh 5% of battery 1,101 mW Drain 0% HW: 0%
    2 18:41:46 7:00:05 9,107 mWh 22% of battery 1,301 mW Drain 0% HW: 0%
    3 2014-10-05<br>13:37:24 0:39:48 759 mWh 2% of battery 1,147 mW Drain 0% HW: 0%
    4 15:18:28 0:48:24 - - - Charge 0% HW: 0%
    5 16:08:40 2:11:25 - - - Charge 39% HW: 37%
    6 22:31:13 8:20:23 2,559 mWh 6% of battery 307 mW Drain 92% HW: 92%
    7 2014-10-06<br>06:55:15 2:30:11 948 mWh 2% of battery 379 mW Drain 89% HW: 89%
    8 09:31:05 0:14:31 288 mWh 1% of battery 1,200 mW Drain 61% HW: 61%
    9 10:11:15 1:08:59 561 mWh 1% of battery 489 mW Drain 88% HW: 88%
    10 11:21:15 0:55:58 1,595 mWh 4% of battery 1,713 mW Drain 67% HW: 67%
    11 13:28:10 3:17:01 1,129 mWh 3% of battery 344 mW Drain 92% HW: 92%
    12 16:46:03 16:25:34 5,830 mWh 14% of battery 355 mW Drain 96% HW: 96%

    So what I noticed in this sleep study report is that you are probably not running the factory load of Windows 8.1 on the Surface.  We don't run the factory load either and we're having battery drain issues as well that I've attributed to Cisco AnyConnect.  However what we are also missing in our corporate image is the hibernate after 4 hour setting that's in the factory load.  I can tell from your sleep study that you don't have it either because if you did, your longest Connected Sleep sessions would be 4 hours long.  Once we implemented that, we went from around a 45-50% battery drain overnight to about 15-18%.  Not perfect, but an improvement.  To get further improvements, we'll need to get Cisco to address it in their software. 

    There is a hotfix (KB2955769) you need to install and the registry settings are mentioned earlier in this thread.

    Was this answer helpful?

    0 comments