Why does my SB system use 2x more Watts than in SPCR review?

Got a shopping cart of parts that you want opinions on? Get advice from members on your planned or existing system (or upgrade).

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
DaRuSsIaMaN
Posts: 51
Joined: Mon Dec 28, 2009 10:37 pm
Location: Arizona, USA

Why does my SB system use 2x more Watts than in SPCR review?

Post by DaRuSsIaMaN » Sun Apr 17, 2011 12:03 am

My main components are identical to one of the setups in the review here:
Sandy Bridge, Part 2: Intel DH67BL & Asus P8H67-M EVO H67 Motherboards

I am using the P8H67-M Evo motherboard with a SB 2500k. According to the review, the test system ran at 20W at idle (linked above). I am having the computer sitting idle at the desktop of my newly installed version of Win7, and measuring power consumption at the outlet using a P3 Kill A Watt: I'm seeing 38W.

That is disappointing. From what I understand, there should be a difference between the estimated DC power given in the review versus measuring it directly at the outlet as I'm doing. Is that right? But, I would not expect that my way of measuring would give a result of twice as much! What's going on here?

Here are the full specs of my system:
i5 2500k
ASUS P8H67-M Evo (Rev 3.0)
Kingston SSDNow V+100 96gb ssd (Windows 7 Professional, 64 bit is installed on this drive)
WD10EARS (1gb wd caviar green drive) also plugged into mobo... not even formatted, still raw
stock cpu cooler
4gb (2x2gb) avexir 1333mhz, 1.50V RAM
asus SDRW-08D1S-U external dvd-rw drive connected via USB
Seasonic S12II Bronze 380W PSU
My case has 3 120mm fans, pretty low rpm
Have not installed any intel graphics drivers yet, nor even the ASUS motherboard drivers/utilities that come on CD with motherboard... have not really done anything beyond fresh install of Windows (not sure if this might mean anything, but the specs page in the review mentions the intel graphics driver version)

Any ideas? And has anyone else with any sandy bridge build tried measuring Watts at the outlet, as I'm doing? I'd like to hear what you get. Even if your setup is significantly different. Thanks!

jhhoffma
Posts: 2131
Joined: Mon Apr 25, 2005 10:00 am
Location: Grand Rapids, MI

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by jhhoffma » Sun Apr 17, 2011 7:40 am

Doesn't take a lot to add 18W to a system. 1/3 of that could be your second HDD (WD Green). You also have 3 fans in your setup (1-2W each).

Not to mention your PSU is a tad less efficient than the one SPCR uses.

Combined with the potential inaccuracy of a Kill-A-Watt, I think you're doing just fine...

Might want to check to see if you've enabled the power saving functions in the BIOS, and you can always undervolt...

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by HFat » Sun Apr 17, 2011 8:22 am

I don't understand the OP's surprise. You thought you only had to use a CPU with low idle power consumption to get low system power consumption?
I've said it several times on these forums and some people don't want to listen but if your CPU's TDP forces you to get a PSU that's inefficient at low DC output, that's what you must expect. And that's why you can lower consumption without overspending by going with the low-wattage CPUs, not because they idle lower.
Unsurprisingly the OP's PSU is inefficient. Assuming his measurement is accurate, the system is pulling what, 26W DC? Maybe 27. 4 or 5W could be explained by some of the factors jhhoffma highlighted (not 18W... the OP was comparing AC and DC!). I haven't compared SPCR's and the OP's setup so I don't know.

ignatz
Posts: 19
Joined: Mon Apr 13, 2009 3:36 am
Location: Way Out

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by ignatz » Sun Apr 17, 2011 9:44 am

Per Kill-a-watt, I use about 75 watts at idle with these components:

1 i-5 2500 CPU, using integrated graphics
1 Gigabyte H67A-UD3H-B3 motherboard
1 WD Caviar Black drive
1 WD Green drive
4 GB Crucial RAM (2 sticks of 2 GB each)
1 Seasonic S12-330 power supply
1 Samsung DVD drive
1 NEC 19 inch LCD monitor
1 Scythe Slipstream fan, 800 rpm
1 Scythe Slipstream fan, 500 rpm
1 Intel retail heatsink and fan
1 Motorola cable modem
1 HP PS 2 keyboard
1 Microsoft optical mouse


This compares to 115 watts at idle with a near identical system running a Core 2 Duo E 6600.

The reduction from 115 watts to 75 watts saves me about $18 per year in electricity figuring 4000 hours per year of use.

Power usage under load (Prime 95) per Kill-a-watt on both systems is near identical at 160 watts.

DaRuSsIaMaN
Posts: 51
Joined: Mon Dec 28, 2009 10:37 pm
Location: Arizona, USA

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by DaRuSsIaMaN » Sun Apr 17, 2011 6:26 pm

jhhoffma and HFat, why is it that you guys think my PSU is less efficient than the one used in the SPCR review? It looks to me that they would be either identical or mine might even be marginally better. They're both rated at 80plus bronze, but mine is slightly lower wattage. (380W vs. 400W)
HFat wrote:Assuming his measurement is accurate, the system is pulling what, 26W DC? Maybe 27.
Can you or anyone else explain how the conversion works? I'm sure I could understand it if I see an explanation, but I don't know how you convert from AC to DC.
HFat wrote: I haven't compared SPCR's and the OP's setup so I don't know.
Well, I linked the spec page from the SPCR review in my OP, but here is the link again. But, to make things easier, I copied/pasted it below:
Test Setup:

* Intel Core i3-2100 processor - 3.1GHz, 32nm, 65W, integrated GMA HD 2000 graphics
* Intel Core i5-2500K processor - 3.3GHz, 32nm, 95W, integrated GMA HD 3000 graphics
* Scythe Kabuto CPU cooler
* OCZ Platinum Extreme Low Voltage memory - 2x2GB, DDR3-1333
* Western Digital Scorpio Blue notebook hard drive - 500GB, 5400RPM, 8MB cache
* Asus BC-1205PT Blu-ray drive - SATA
* Seasonic SS-400ET ATX power supply
* Microsoft Windows 7 operating system - Ultimate, 64-bit
* Intel GMA 15.21 graphics driver
***

Thanks for the info, ignatz. Just need a little clarification. Do you have the PC itself AND the monitor plugged into Kill-A-Watt? And modem too? In other words, you have them plugged into a power strip and then measuring watts from the entire power strip? I am measuring the PC itself only, because that's how SPCR did it. That is, the power cord from my PC goes into Kill-A-Watt, and nothing else.

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by HFat » Sun Apr 17, 2011 6:59 pm

I did not try to model your PSU. I just guessed based on SPCR's table in the PSU review. Look at it and take your own guess:
http://www.silentpcreview.com/article753-page4.html
Actually I made a mistake in my previous post. I had guessed 24 or 25W (as you can see from the rest of my comment where I speak of 4 or 5W in excess of 20W) and then counted from 22 instead of 20.

I don't know which PSU is the most efficient. What matters is that:
-they are both high-wattage PSU so their rating (bronze, gold and so on) is irrelevant when the system pulls 5-7% of the PSU's capacity
-they are both inefficient when the system idles (as opposed to a pico which is designed for that type of loads)
-the 20W was DC and you are measuring AC so the inefficiency of the PSU makes a big difference

Going from a notebook HD to a desktop HD and a SSD is going to increase power consumption quite a bit. Your actual idle power consumption could well be higher than 38W.
Maybe you can reduce your power consumption with the graphics driver or something but that would mean your measurement was more inaccurate than we think or that your sample is more efficient than SPCR's, not that you bridged part of the gap with SPCR's 20W. That's impossible without changing the PSU.
Last edited by HFat on Sun Apr 17, 2011 7:05 pm, edited 1 time in total.

ignatz
Posts: 19
Joined: Mon Apr 13, 2009 3:36 am
Location: Way Out

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by ignatz » Sun Apr 17, 2011 7:05 pm

Russia Man:

The figures I quoted are for ALL of those components, using a 330 watt Seasonic. Everything is plugged into a power strip and I plugged the power strip into the Killawatt.

I did some more experimenting today and can tell you that the modem and monitor combined take about 28 watts.

Also, I switched power supplies today to a Seasonic SS-560-KM. With it, my max power usage from the wall was 144 watts under the same load that hit 160 watts using the older power supply. Power consumption at idle dropped slightly to 69 watts from the wall per Killawatt.

Running Prime 95 "in place large" today (with the new PSU), HWInfo32 showed "CPU package power" peaking at 63 watts and idling at between 4 and 5 watts. These are the same figures I reached using a Core 2 Duo E6600 and the old power supply.

Oh--also today's tests used a Scythe Big Shuriken heatsink with a slow 120mm fan rather than the stock Intel retail fan mentioned in my first post.

I'm not sure what it all means, but you can probably make some inferences.

m1st
Posts: 132
Joined: Sun Jan 31, 2010 6:43 pm
Location: US

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by m1st » Sun Apr 17, 2011 8:51 pm

I just perused the thread so sorry if this has already been answered...

DaRussiaMan, the reason why your power consumption seems so high compared to the review is because you are looking at two seperate figures. You see 38w, but that's the AC power from the wall. The review quoted estimated DC power consumption, which takes into account the inefficiency of the power supply at that load. I'm doing the math as I write this so hopefully I'm right :-p

The lowest point of measure on for your power supply from the S12II-380 review was at 21.6w DC, which translates into 33w AC. This puts the calculated efficiency for the PS at 65% at that load. We'll use that as a baseline. Keep in mind that we're extrapolating, but it should be close enough...

You said that your computer drew 38w from the wall. Multiply that by .65, and you get 24.7w. So it looks like your rig draws 24.7w. As to what is causing the 4w descrepancy, it could be a random setting, sample variance, testing variance, or a myriad of other things.

Time to re-read the thread to see if this is a repeat :-/

**Edit**

Good, no repeat. If you were still interested how I got the 65% efficiency number, click on the link above regarding the Seasonic power supply test. The first table on the page lists what SPCR found the efficiency of the power supply to be at all those points. It may be shocking to see that an 80 Plus Bronze power supply can have such a low efficiency, but that's how it is. Typically power supplies hit their best efficiencies between 50% and 80%. If it's any consolation, at the low wattage levels your computer is idling, the energy lost to the PSU inefficiency is barely enough to change my phone, so I wouldn't worry too much about it :).

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by quest_for_silence » Mon Apr 18, 2011 12:30 am

DaRuSsIaMaN wrote:Can you or anyone else explain how the conversion works? I'm sure I could understand it if I see an explanation, but I don't know how you convert from AC to DC.

Not to mention Killawatt accuracy, samples variance, or any possible measuring errors, IMO a ballpark calculation is still doable, even if we haven't comprehensive data about the S12II Bronze efficiency (particularly at those low power levels).

If we give a look to 80plus/ECOS certification (from Oklahoma Wolf on JonnyGuru and also SPCR findings we know that often they are already a bit optimistic), we have an 83.88% efficiency at the first test point (75W DC/89W AC) for your PSU.

Interpolating and averaging the previous S12-II 80+ SPCR findings, we should have a predicted efficiency at the same level of from 80.24% up to 81.83% for this latter: let's average and say 81.04%.
If so the difference between the 80+ and the 80 Bronze S12II could (should) be about 2.84% (at that 75W DC level).

Assuming that relative efficiencies may scale about linearly, providing that the ~21.6W DC (about the 5.7% of nominal max load) efficiency of the S12II 80+ was measured at about 65.4%, and eventually adding the previous estimated delta, the S12II Bronze efficiency should be NOT LESS than 68.2%.

Then we have a further data from Jonny Guru: the S12II Bronze (but the 520W one) was measured at 82.3% at 50W DC, about the 9.6% of that PSU nominal max load (pretty impressive): surely we also know that the efficiency ratio at a 10% load easily can be very different from the one at 5%, but IMHO we may try to do some somehow reasonable guesswork about the upper limit of a possible, predicted efficiency of your S12II Bronze at its 5% rated load level.

From the old SPCR review, we also know that the S12II 80+ 11.2% rated load level efficiency is a mere 76.3%.
The difference between the previous JG's finding (even if the load levels are slightly different) is a full 6%.
Therefore I assume that the S12II Bronze efficiency should be NOT MORE than 71.4% at around the 5% rated load level (20-22W DC).

Furtherly rounding by averaging those supposed minimum and maximum foreseable efficiencies, we can eventually suppose a nearly 70% efficiency at ~20W DC for your S12II Bronze 380W PSU.

So, if it were true, how many DC Watts do your measured AC watts match? 38W AC * .70 = 26.6W DC.
Then you have an about (at least) 1.85W DC penalty for your WD Green drive (over the Scorpio Blue), and an about (at least) 4W DC penalty for your case fans, for a total amount of aboutt 5.9W DC more power consumption.

Finally, 26.6W DC minus 5.9W DC give 20.7W DC: which I think it might be fairly in line with SPCR finding (20W DC), not to mention Killawatt accuracy, samples variance, or any possible measuring errors. What do you think?

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by HFat » Mon Apr 18, 2011 2:47 am

There is also the SSD which consumes a little bit. And less than 2W for the 3.5'' drive seems dubious. On the other hand, 4W for the fans may be excessive. And the consumption of the RAM may be a bit different. Whatever. The only large values are the board's idle power consumption and the PSU efficency anyway.

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by quest_for_silence » Mon Apr 18, 2011 3:56 am

HFat wrote:There is also the SSD which consumes a little bit. And less than 2W for the 3.5'' drive seems dubious. On the other hand, 4W for the fans may be excessive.

Perhaps his SSD (I falsely assumed even the SPCR test system sported one) may account for that residual 0.7/0.8W (or whatever) DC difference over the SPCR findings.

On the contrary, the 1.85W is the actual difference between the WD Scorpio Blue 500GB and WD Caviar Green 2TB idle power consumptions: according to SPCR, the former goes around 0.95W, while the latter should draw 2.8W (hence the less than 2W difference), both with heads unlocked. Providing it has the WD Caviar Green 1TB (less platters, IIRC), it could (should) be even less than that.

About those possibly excessive 4W, thinking it could have been an acceptable trade-off, I've put into account about 0.11A for each case fan (which should be three, IIRC: a 1200rpm fan like a Slipstream usually draws 0.30/0.40A at full speed), providing that the OP hasn't pointed out the relevant specs.

Eventually, AFAIK idle power consumption and PSU efficiency (apart sample variance) are quite the same for both his and SPCR systems.

PaulShapiro
Posts: 24
Joined: Sat Jun 07, 2003 9:30 am
Location: New York City

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by PaulShapiro » Mon Apr 18, 2011 5:10 am

In the initial post you mentioned that you had not yet installed the Intel chipset drivers. It's possible that optimal cpu power saving only occurs after you install the chipset drivers, so I would try that and see if anything changes. As previous posters noted, you're only looking to explain a few watts difference. I can tell you that according to FinalWire's Aida64 system monitoring software, my 2600K is using about 4-5 watts at idle. Under full load it goes as high as 80 watts, so any difference in cpu management could have a noticeable effect.

DaRuSsIaMaN
Posts: 51
Joined: Mon Dec 28, 2009 10:37 pm
Location: Arizona, USA

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by DaRuSsIaMaN » Mon Apr 18, 2011 8:53 pm

Awesome, thanks for all the comments regarding power supply stuff, and special thanks to quest_for_silence for the detailed analysis. So, now I see that the conversion between AC and DC watts is extremely straightforward -- just different by the factor of efficiency. I was thinking there was something more involved it due to the large discrepancies. Basically, I did not expect the efficiency to drop so low even though I knew that the 80plus certification stuff only applies to like 20% load at the bottom end of the rating. Anyway...

1.) Regarding case fans. First, question: when SPCR did that review, can we assume that they had no case fans? Were they using a test bench or something, rather than a real case? Can we assume this because they did not report any case fans in the spec sheet? I honestly figured that the reviewer just didn't bother mentioning the case and case fans.

Back to me. It looks like my fans are each about 1W (seen at the outlet by Kill A Watt), at least when they're at the setting I had them at, which was low. I think I actually should have reported a slightly higher measurement in my original post, like 39-40W. I remember it was showing that measurement, then the computer sat around for a while, and Kill A Watt started showing 38ish watts. Strange. So I just reported that in my first post. (It was not in sleep/standby but it's possible that the monitor had gone to sleep.)

So let's use my new baseline. I have by now installed all sorts of drivers. Intel display driver, the intel chipset drivers, Rapid Storage Technology driver, network driver... I think that's it. With this new status quo, and EPU set to Auto, and with all case fans completely unplugged, I now see 36W. (It actually flips back and forth between 36 and 37, with some random spikes going higher.)


2.) Regarding Kill A Watt inaccuracy. I think we can assume that it overestimates the Watts by ~1-2W at this range. When nothing is plugged into it at all, it shows 3W (not zero). lol. When I plugged in a 14W-rated desk lamp (has a CFL) into it, it showed 17W. However, this offset doesn't seem to last as you go higher. I plugged in a 25W desk lamp (some oddball halogen bulb); it showed 27W. I plugged in a 40W desk lamp (another halogen); it showed 41W.


3.) Found the main culprit: HDD. So, after establishing the new baseline -- with drivers installed and all case fans unplugged -- at 36W, I then went to test the WD10EARS HDD. I unplugged it. Powered on, and now the Kill A Watt reading flips between 30 and 31W (spends more time at 31 though). So, that's an entire ~6W difference just for eliminating the HDD. Now, if I further subtract 1 or 2W from this reading to account for Kill A Watt inaccuracy, that gives say 29W. 29W * 0.7 = 20.3W DC. So that now indeed falls entirely in line with SPCR review.

I'm kind of annoyed, though. I expected that HDD to make like 2W of difference. SPCR shows 2.8W idle for the 2TB model, and Tom's Hardware reports specifically the 1tb model (the one I have) at 2.8W idle. See here. Soo ... why is it registering like 6W for me? Again, using assumed efficiency of 0.7, that translates to 4.2W DC, 50% higher. I should probably quit worrying about a few Watts lol...

DaRuSsIaMaN
Posts: 51
Joined: Mon Dec 28, 2009 10:37 pm
Location: Arizona, USA

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by DaRuSsIaMaN » Tue Apr 19, 2011 11:07 pm

So, does anyone know the answer to this?
DaRuSsIaMaN wrote: 1.) Regarding case fans. First, question: when SPCR did that review, can we assume that they had no case fans? Were they using a test bench or something, rather than a real case? Can we assume this because they did not report any case fans in the spec sheet? I honestly figured that the reviewer just didn't bother mentioning the case and case fans.

Rebellious
Posts: 322
Joined: Sun Dec 09, 2007 12:53 pm
Location: EU, USA

Re: Why does my SB system use 2x more Watts than in SPCR rev

Post by Rebellious » Fri Apr 22, 2011 5:23 pm

38W is about average for today's systems, Intel or AMD. I get 39.5, see 2nd post with photo in thread:

viewtopic.php?f=28&t=60616

Post Reply