Re: Windows 10 and 11 power state habits: shutdown, sleep or hibernate

On Mon, 1/26/2026 8:59 AM, Frank Slootweg wrote:
> Steve Hayes <[email protected]> wrote:
> [...]
>> I shut down my desktop and laptop at least once a day, when I go to
>> bed at night, and I switch them off at the wall plug, mainly to keep
>> my electricity bill down. Even when the machines themselves are not
>> running, the transformers consume electricity if not switched off or
>> unplugged. 
> 
>   If the system sleeps, it should use very little power, only for
> maintaining the content of RAM.
> 
>   If the system is hibernated, it uses even less or no power. Paul can
> probably tell for a desktop which part(s), if any, still use power. The
> laptop does not use any power, which you can prove by taking out the
> battery (if that can be (somewhat easily) removed. After re-inserting
> the battery (after a couple of hours) and shorthy pressing the power
> button, the laptop will resume where you left off.
> 
>> My desktop computer has a couple of hard disks, and I have a funny
>> feeling that if I leave it running all the time, even when I'm not
>> using it, the bearings will wear out faster. 
> 
>   Both with sleep and hibernate the disks should stop.
> 
>   For details, see the settings in the Power Options applet (Control
> Panel -> Power Options -> Change plan settings -> Change advanced power
> settings).
> 
>   Bottom line: Do as you please, but don't do things a certain way for
> the wrong reasons, i.e. in this case (no) electicity use and (no) disk
> wear.
> 

If the PC is on at the back but not by definition doing
anything, the power consumption just from the PSU running
is 1 watt or so. But if you use your Kill-O-Watt meter and
plug the PC into the meter, the power factor is terrible.
You might find the reactive power (something you are not
billed for) is 7VA and the real (bill-able) power is 1W.
It appears there is no power factor correction at
that point.

The sleeping computer, you could count 1 watt per DIMM,
as a rough estimate.

If you have Wake On LAN enabled (say, from Device Manager),
the motherboard will leave the NIC chip running, and
that costs you (approximately) 1 watt. It's also possible
to configure some HID (mouse/keyboard) for system wake functions
and then the USB ports are powered, and that might cost
another 1 watt.

My networking equipment (which runs all the time), it
draws 17 watts, and that can be more than the PCs
happen to be drawing when in one of the reduced states.

If I "flip all the switches in the room", the room falls to 0.0 watts :-)
But then my VOIP phone would not work. No incoming calls.

The light bulb is maybe 13 watts (LED).

Using a Kill-O-Watt meter, you can measure real and reactive
power. And it is good down to the watt level.

*******

One of the web sites that estimates PC power consumption,
had DIMMs listed as "25 watts each". Which is wishful
thinking, and if that were true, the metal cooling surface
is not rated for 25 watts. The surface temperature would
be quite high if the consumption was 25 watts.

The consumption is a lot lower than that. Each cycle type
(read, write, refresh, NOP) has a different power dissipation,
and we do not assume "100% write" because DRAM cannot do that.
There might be a write-burst-of-four in a one hundred cycle
timing window. During a No-Operation, the clock is running,
but no command is being fed over the command bus, and
so the interface is pretty quiet at that point. It is the
averaging of all these cycle types that gives an overall
power usage.

When a DIMM is sleeping, no commands are being sent to it,
and it is left in Auto-Refresh state, to increment an
internal counter and deliver refresh cycles to a row at
a time. And the refresh cycle doesn't dissipate as much
power as a read or write, and it doesn't happen that often.
But on a laptop and a laptop battery, that dissipation
eventually drains the battery pack. The load isn't zero.

   Paul