Shutdown vs sleep power consumption

Recently completed a new build with Asus Strix z390, i9900k cpu and Seasonic titanium psu so thought how much power it consumes, so used a watt meter at the socket to measure the power draw and found something interesting,
Shutting down the Windows (but not disconnecting power) draws 3w occasionally touching 4w (USB power disabled, LEDs disable, but not EuR power saving ebabled as that makes computer do double boot).
Sleep mode - 4w

with my old build this used to be 3w off and around 10w sleep so looks like improvements to power savings in the CPU and memory have made massive gains. With the benefit of not having to close anything and instant start, I am going to use Sleep button from now. :)

Interesting to see there isn't much difference in usage. Do you know what's still drawing power when shutdown?

I'd be interested to see the difference with EuP on!

I wonder the same, all LEDs are off and USB power is set to off. Didn't try with the EuP as that force the double boot during power on, but I will try that and report what I find.

The accuracy of consumer mains power meters is iffy when it comes to low power figures like single digit Watts.

Such that, I wouldn't have any confidence making comparisons between shut down vs. sleep power consumption if we're looking at 3-4W out of one of those units.

On topic, I'd imagine much of the power consumption would be down to keeping the power manage circuits alive - be it on soft-shut down or standby, and given the tiny power consumption the power efficiency would be terrible. I don't know how much power it would take keeping the system RAM alive for standby though... I'd guess for DDR4 it should be fairly low.

Check out this PSU efficiency graph;

https://seasonic.com/pub/media/wysiw...s/Titanium.png

The graph falls off sharply to the left, you can imagine how much the efficiency falls off when there is a fraction of a percentage load!

Have you turned wake on Ian off. If not the networ port may still be on when shutdown.

Quote:

---

Originally Posted by DDY

The accuracy of consumer mains power meters is iffy when it comes to low power figures like single digit Watts.

---

That is true, so the 4W figure could be 2W or 8W, but the fact that the measurement is the same should be reliable. ie the accuracy might be rubbish, but the precision (repeatability) should be OK.

I did a similar test some years ago with a cheap consumer watt meter, and as a quick sanity check I plugged in a desk lamp with a 7W LED bulb in it (that was the sort of standby power I was seeing on that PC). The reading was surprisingly close.

Quote:

---

Originally Posted by DanceswithUnix

That is true, so the 4W figure could be 2W or 8W, but the fact that the measurement is the same should be reliable. ie the accuracy might be rubbish, but the precision (repeatability) should be OK.

---

I disagree.

The error delta purely on rounding between a 3W and 4W reading could be as much as 1.99W which is a massive error (4W could be 4.49W max and 3W could be 2.50W min), that's 67% of 3W!

Where as, between say 9W and 10W the max absolute error would be still be 1.99W, but as a percentage of the lowest reading is now 22%, which is still a bit iffy.

The readings could be bang on 3.000W and 4.000W for all we know - it's iffy.

Quote:

---

Originally Posted by banana999

Interesting to see there isn't much difference in usage. Do you know what's still drawing power when shutdown?

---

I'll hazard a guess that there is still some power to wake/remote-start the PC. After all, that's how our IT guys at work can remotely apply updates across the company at 3am, even if we shut down our machines fully. Only way thay can't is if we power off the PSU.

Also, you'll have a small draw for things like keeping the CMOS battery charged, and to any peripherals you have connected. I also reckon there'll be a small current for USB passthrough, so you can still charge your phone even if the PC is off.

Just my guess, of course, but I'll bet a lemon drizzle cake there's at least some truth in it!! :)

Quote:

---

Originally Posted by DDY

The readings could be bang on 3.000W and 4.000W for all we know - it's iffy.

---

Yeah, I know all about significant figures in measurements. But this is a value call the OP is making.

We don't know the offset or non linearity of the meter, but in this case used in a comparison between two very similar readings we can be fairly certain the difference is under 2W which for some will be worth spending for a faster startup time. Heck, I've known people not ever switch off their PC at an expense of a steady 60W so it is always instantly available.

If you could say "actually that might be a 10W difference" then that would be different. But I don't think you can.

Edit: Oh and the PSU efficiency thing, if it is measuring 4W at the plug that just means it is using 2W in main PC, which is kind of interesting but not really relevant here.

Quote:

---

Originally Posted by DanceswithUnix

… Edit: Oh and the PSU efficiency thing, if it is measuring 4W at the plug that just means it is using 2W in main PC, which is kind of interesting but not really relevant here.

---

Well, in terms of the OP's decision it's not relevant, but in terms of banana's follow up question on what's drawing power it's a very significant difference. I can totally see 2W of standby draw across the 5VSB and 12VSB (does anything even use that anymore?) rails...

Quote:

---

Originally Posted by DanceswithUnix

Edit: Oh and the PSU efficiency thing, if it is measuring 4W at the plug that just means it is using 2W in main PC, which is kind of interesting but not really relevant here.

---

I think efficiency is worth looking in to even in this context because it can plausibly account for a significant difference between the soft-off and sleep state.

To re-iterate, my point is that the closer to zero load the steeper the efficiency curve falls off - hence the potential for significant differences when comparing the two states which are apparently separated by 0.5 - 1.5W.

Do you have a Haswell PSU compatibility option in your BIOS? It loads your PSU in sleep to avoid the PSU shutting down the motherboard as Haswell chips went into a much deeper sleep than earlier CPUs.

I can't measure a difference between sleep and off on my computer as my metre only measures whole watts, so it says 1w for both.

Yes, certainly agree that the watt metre I got is not very accurate, but it was showing a consistent reading of 4w for sleep or 3w for shutdown or goes to 0 when off with power switch on psu.

The psu Seasonic titanium, has a relay in it and I can hear that clicking for both shutdown and sleep so it must be isolating mains for larger part of the circuit to keep efficiency high, so I guess the standby power section that works under both conditions must be drawing similar power in both situations.

I always choose to shut down and off at the plug. fastest for me..