Setup pulling 900W from the wall?

[done12many2]

10 minutes ago, ShadowWolf810 said:

So I mean there's a considerable overclock in there, but would the overclock and accessories account for a 400 W difference between your result and what reviews post as the system under load?

 

Unless you know exactly what it is that the reviewers are sharing, you shouldn't compare your numbers to theirs. 

 

Some measure at the wall while others measure current at the 12v EPS.  In addition to those variations, you also have great variations in load types. Contrary to popular belief, 100% CPU usage is NOT a measurement of load.  Run AIDA64 stress test and record the power draw at 100% CPU usage and then do the same thing while running the latest Prime95 with AVX at 100% usage.  I promise you you'll see a HUGE difference in not only temps, but power drawn at the wall. 

 

Your numbers are right where they should be. Enjoy it. 

[Sakkura]

59 minutes ago, Bananasplit_00 said:

50W for the rest seems quite low, its an X99 board, 240mm AIO, at least one harddrive and probably one SSD. a harddrive is at least 10W, SSDs more

Are you kidding me?

 

50W is, if anything, a bit on the high side.

 

And no, SSDs do not use more power than HDDs.

[ShadowWolf810]

4 minutes ago, done12many2 said:

 

Unless you know exactly what it is that the reviewers are sharing, you shouldn't compare your numbers to theirs. 

 

Some measure at the wall while others measure current at the 12v EPS.  In addition to those variations, you also have great variations in load types. Contrary to popular belief, 100% CPU usage is NOT a measurement of load.  Run AIDA64 stress test and record the power draw at 100% CPU usage and then do the same thing while running the latest Prime95 with AVX at 100% usage.  I promise you you'll see a HUGE difference in not only temps, but power drawn at the wall. 

 

Your numbers are right where they should be. Enjoy it. 

Well I know I've seen them specifically use similar wall outlet wattage meters, and was just guessing ballpark numbers, or even they'd test higher end systems than mine, it seemed like 600 W at the wall was just too much when they're getting like 300 ish with Threadripper etc. But if everything seems normal then I won't worry about it. 

 

I am still concerned with the USB errors and just occasional bugginess I experience though. See my reply to Tabs right before this. It still seems like its some sort of shorting issue, or dirty power? 

[done12many2]

12 minutes ago, ShadowWolf810 said:

Well I know I've seen them specifically use similar wall outlet wattage meters, and was just guessing ballpark numbers, or even they'd test higher end systems than mine, it seemed like 600 W at the wall was just too much when they're getting like 300 ish with Threadripper etc. But if everything seems normal then I won't worry about it. 

 

I am still concerned with the USB errors and just occasional bugginess I experience though. See my reply to Tabs right before this. It still seems like its some sort of shorting issue, or dirty power? 

 

You're doing an apple to orange comparisons.  You're comparing your total system power with CPU & GPU in full load, which is not the same thing as comparing total system power of Threadripper with CPU only in load.  I promise you that Threadripper with a full load combined with a 1080 Ti with full load will pull more than 300w from the wall in similar conditions.  

[ShadowWolf810]

11 minutes ago, done12many2 said:

 

You're doing an apple to orange comparisons.  You're comparing your total system power with CPU & GPU in full load, which is not the same thing as comparing total system power of Threadripper with CPU only in load.  I promise you that Threadripper with a full load combined with a 1080 Ti with full load will pull more than 600w from the wall in similar conditions.  

So this is what made me think of posting about all this, from todays LTT video. 

 

 

I guess its not clear what "synthetic load" they're using here so that would account for some difference (and if we're looking at Threadripper still) would the 1080 Ti make up the extra 400 W you're talking about?

 

So referring back to my original results, I had 500 W draw with CPU only, not overclocked. Yes there are a few other peripherals there and two additional monitors, but it LOOKS like those additional things are only accounting for a little bit. 

 

I guess it comes down to trying to estimate 100 variables in your head without super hard numbers and inconsistent testing parameters

[Tabs]

4 hours ago, ShadowWolf810 said:

So this is what made me think of posting about all this, from todays LTT video. 

 

 

I guess its not clear what "synthetic load" they're using here so that would account for some difference (and if we're looking at Threadripper still) would the 1080 Ti make up the extra 400 W you're talking about?

 

So referring back to my original results, I had 500 W draw with CPU only, not overclocked. Yes there are a few other peripherals there and two additional monitors, but it LOOKS like those additional things are only accounting for a little bit. 

 

I guess it comes down to trying to estimate 100 variables in your head without super hard numbers and inconsistent testing parameters

Before I say what I was going to say, I clicked that and ended up in the depths of YouTube for 4 hours. So take that as you will.

 

Short answer: Yes, that makes your system appear far less efficient. However, a server motherboard - even one with two chipsets - is intrinsically designed for efficiency. Each of the *four* Opteron 6276 processors they tested have 115W TDP, versus the 140 of your cpu at stock. 

 

When LTT (and everyone in the industry) tests CPU power consumption, they don't put load on the GPU as well. The point of the task is to pull watts from the CPU. The wattage listed there is from pushing all 4 CPU's as hard as the system can. Your high wattage result was from pushing both an (overclocked and overvolted) CPU and your GPU at once.

 

So now, to put that value into perspective. The Threadripper cpu benchmarked there with power consumption figures was 217W. That would be the full system - without GPU load or monitor. If you add 250W for your GPU and 40W for your monitor (or remove those from your current measurement), and *then* remove the extra power added by your overclock, what figure do you get for your machine?

 

I'm guessing probably in the same ballpark or lower.

 

Seriously, your machines power draw is completely fine for what you're doing with it. If you want it to be more efficient, drop it to stock volts and run it as high it can go at that vCore. Since power consumption is linear with frequency and exponential with voltage, the higher you can go at stock/reduced volts, the more efficient your overall machine is. 

[Bananasplit_00]

10 hours ago, Sakkura said:

And no, SSDs do not use more power than HDDs.

my bad, clearly. they seem to be in the handfull of watts rather then the 10s. 

10 hours ago, Sakkura said:

Are you kidding me?

 

50W is, if anything, a bit on the high side.

he has 64GB of RAM, a board that draws those 50W by itself, two harddrives, and 5 fans. thats not fitting in those 50W no matter what way i look at this when the fans alone draw half what you estimate

[paddy-stone]

33 minutes ago, Bananasplit_00 said:

my bad, clearly. they seem to be in the handfull of watts rather then the 10s. 

 

Yeah IIRC, SSDs typically consume approx <3W - HDDs vary quite a bit, but most newer HDDs, let's take a WD red drive for example, is around 7W (rounded up), taken from HERE

 

 

This isn't a definitive collection, but from what I can recall that seems about right in line with my experiences with SSDs and HDDs. Desktop HDDs typically consume between 6-12W, with drives with more platters and/or higher RPM usually being the higher of the two figures.

 

OP, from looking through this thread I mostly come to the conclusion that the static issues you're having seem related to a short somewhere, most likely caused by USB devices or ports... If I were you I would try to whittle your USB devices down to the lowest possible, and thoroughly test to see if it still does it. Also try different ports and/or USB hubs, and I know they don't look great, but powered USB hubs can sometimes be a blessing. That might help you figure out where the short is coming from by trying the different USB ports one at a time with the USB hub.. I suspect it's to do with your I/O ports, but could be wrong... the reason I say this is because of the flickering you get on your monitors, so could be the static discharging across your display ports maybe. I'm not an electrician, but that makes sense to me.

[Sakkura]

5 hours ago, Bananasplit_00 said:

he has 64GB of RAM, a board that draws those 50W by itself, two harddrives, and 5 fans. thats not fitting in those 50W no matter what way i look at this when the fans alone draw half what you estimate

The board does not draw 50W by itself, that would be ludicrously inefficient. And the hard drives will be mostly idle.

4 hours ago, paddy-stone said:

Yeah IIRC, SSDs typically consume approx <3W - HDDs vary quite a bit, but most newer HDDs, let's take a WD red drive for example, is around 7W (rounded up), taken from HERE

And those power consumption figures are when they're being hit with a benchmark. OP's situation would have them drawing a small fraction of that wattage.

[Bananasplit_00]

2 minutes ago, Sakkura said:

The board does not draw 50W by itself, that would be ludicrously inefficient. 

8 phase VRM pumping out 215W? i dont see that not loosing at least 30W in MOSFETs, and then we have SATA, audio, NIC, and a chipset which acording to Intel says the chipset draws 6.5W, so sure it might be a high estimate but the memory is going to, now that i have actiually looked into this a bit more, be drawing somewhere around 20W range.

 

so lets go back to that 50W for the rest of the system figure you mentioned, so with just the RAM, chipset and fans we have already gone beyond it, so thats not counting any VRM losses, stuff plugged into the IO and drives. it dosent hold. oh and rendering with resolve would be wrighting data to your disk would it not?

[AshleyAshes]

18 hours ago, ShadowWolf810 said:

All devices "shutdown" but not unplugged : 24 W

With this alone I think the smart move is to sanity check your Kill-A-Watt or whatever else you are using with something much simpler.  Like plug in just a phone charger to it and see how much it reports being drawn.

 

A PC and some monitors, all in stand by, are unlikely to be at more than 5w even.  25w means either the measurement is wrong or you have a serious problem.  So before figuring out what the problem is, ensure that the measurement is accurate.  The measurement itself could be the reason for ALL the other numbers you're seeing.

[paddy-stone]

1 hour ago, Sakkura said:

 

And those power consumption figures are when they're being hit with a benchmark. OP's situation would have them drawing a small fraction of that wattage.

I never said it would be consuming that much, notice the < sign, meaning less than? and it was just a generic answer to an earlier question anyway about how much a hard drive would pull. Obviously if they are on idle then they would consume less, much less in some cases.

Wasn't the original question asking about max power draw anyway? in which case you have to assume he means hard drives in use also.

[Sakkura]

1 hour ago, Bananasplit_00 said:

8 phase VRM pumping out 215W? i dont see that not loosing at least 30W in MOSFETs, and then we have SATA, audio, NIC, and a chipset which acording to Intel says the chipset draws 6.5W, so sure it might be a high estimate but the memory is going to, now that i have actiually looked into this a bit more, be drawing somewhere around 20W range.

 

so lets go back to that 50W for the rest of the system figure you mentioned, so with just the RAM, chipset and fans we have already gone beyond it, so thats not counting any VRM losses, stuff plugged into the IO and drives. it dosent hold. oh and rendering with resolve would be wrighting data to your disk would it not?

The VRMs don't really count for the motherboard. It's determined mostly by the CPU. If the CPU draws another 50W for itself because you overclocked it, that's hardly the motherboard's fault, nor is the extra few watts of VRM losses that go along with it.

 

And yes, the 6.5W chipset is by far the most power-hungry part on the motherboard, showing how extreme that 50W claim is.

[ShadowWolf810]

On 11/6/2017 at 10:48 AM, AshleyAshes said:

With this alone I think the smart move is to sanity check your Kill-A-Watt or whatever else you are using with something much simpler.  Like plug in just a phone charger to it and see how much it reports being drawn.

 

A PC and some monitors, all in stand by, are unlikely to be at more than 5w even.  25w means either the measurement is wrong or you have a serious problem.  So before figuring out what the problem is, ensure that the measurement is accurate.  The measurement itself could be the reason for ALL the other numbers you're seeing.

There's a lot more than just a pc and monitors all in standby though. 

 

Those, as well as an external DAC and Amp both with their own power bricks but those were powered off, a turntable in standby and a phono preamp with no on off switch, Router/Modem was actually on and functioning in that measurement, as well as an external harddrive enclosure with two regular PC drives in it was on (I think). I'm sure probably something else too. It was basically the status that my setup is in when I just wake up in the morning. I'm pretty sure everything draws a small amount of current even if its totally off 

 

It is a Kill-a-Watt device, but it seems to functioning correctly. It reads 0 with nothing connected, and with my phone plugged in as you said it fluctuates between 7 and 11 W. The power adapter says its 5V at 2 Amps, which the math says should be pulling 10 W