How much power does your system draw at idle (Poll)

All about them.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

marcmercer
Posts: 26
Joined: Mon Jul 18, 2005 2:30 am

How much power does your system draw at idle (Poll)

Post by marcmercer » Wed Apr 26, 2006 12:00 am

Having seen Mike mention that one of his systems runs at just 36W idle , I'm curious to hear from everyone else who has the ability to measure how much power their systems use. I'm asking having just built a new htpc, upgrading from a socket A system and not seeing as much of an improvement as i expected, especially as i understood that the 90nm 6150 chipset draws less power than than the stanard 110nm NF4 chipset. Perhaps we can find some trends that show which chipsets draw less power and which PSUs are more efficient at lower draw.


Motherboard: MSI K8NGM2-FID (NF4 90nm)
GFX: On board 6150
CPU: E3 3000+ venice (939) - stock voltage running CnQ
Ram: 2x256mb PC3200
PSU: FSP 300W (fan swapped for nexus @ 7v - saved nearly 3W!!)
Fans: 1xCase fan Nexus @ 7V
Other: 2 x Compro T200 Digital TV Tuners


Total System Draw (measured at wall socket)

Idle: 68W
Load: 106W+

Marc

Mikael
Posts: 206
Joined: Mon Dec 06, 2004 3:12 am
Location: Gothenburg, Sweden

Post by Mikael » Wed Apr 26, 2006 12:44 am

Well, that PSU is supposed to be pretty inefficient. Could be in the lower 60% range for this particular power load. That would mean a low 40W figure for your system components at idle, which is pretty okay.

I don't have the possibility to measure my system at the moment and it's also an old undervolted Socket A machine, so it might not be very interesting.

marcmercer
Posts: 26
Joined: Mon Jul 18, 2005 2:30 am

Post by marcmercer » Wed Apr 26, 2006 1:07 am

yes - I realise the PSU is pretty long in the tooth now and probably isn't at all efficient at low loads, but I'm holding on until the antec fusion case is released :)

Any system power readings would be interesting socket a or otherwise. I should perhaps add my outgoing socket A system:


Motherboard: Abit NF7 v2.0
GFX: Radeon 9550
CPU: Barton 2500+ @ 1.45v
Ram: 2 x 256mb
HD: Spinpoint 160gb
PSU: Enermax 365W (80mm
Fans: 1x120mm nexus @7v 1x92mm Noiseblocker @ 7v
Other: 2 x PCI DVT-B Compro T200 TV tuner cards
1 x PCI sata card

Total system power measured at wall socket:

Idle: 81W
Load: 115W+

openwheelformula1
Posts: 211
Joined: Thu Sep 22, 2005 9:58 am

Post by openwheelformula1 » Wed Apr 26, 2006 1:36 am

idle: 105 watt

Bluefront
*Lifetime Patron*
Posts: 5316
Joined: Sat Jan 18, 2003 2:19 pm
Location: St Louis (county) Missouri USA

Post by Bluefront » Wed Apr 26, 2006 1:58 am

My first SFF system used a K63-333, had embedded everything, used a laptop HD and a laptop optical drive. It had one small 60mm fan on the CPU, and an external wall-wart type power supply of 55W.

So what....

Thomas
Posts: 664
Joined: Mon Jun 20, 2005 11:21 pm
Location: Denmark

Post by Thomas » Wed Apr 26, 2006 2:26 am

A thing to consider, is the precision of the meters for measuring the wall socket wattage....

I also think it's hard to judge how much power the chipset uses, without measuring each component individually. That's probably not the easiest task, with RAM blocks, onboard video, onboard soundcard, onboard netcard... And how much do optical drives, cardreaders etc. actually consume?

Personally, I find it's the efficiency in terms of performance/wattage, there is the interesting issue. As long one have an efficient graphic card, CPU and PSU and the overall wattage is low, then it's good. I dont think picknicking about wattage of RAM blocks and chipsets, will change the overall picture much.

marcmercer
Posts: 26
Joined: Mon Jul 18, 2005 2:30 am

Post by marcmercer » Wed Apr 26, 2006 2:59 am

I agree there is the issue of power meter precision and this thread won't give us a definitive x chipset uses more power than y .... but I think that if enough people post it, any trends may give anyone looking to build a particularly low power machine an idea of which components are good and show up any suprisingly power hungry components.

Now that I have anywhere between 3 and 5 pcs running 24/7 in the house, reducing the power consumption by say 10% can have a dramatic effect on the electric bill. There is also the added benefit of being able to cool the system more easily and therefore more quietly. As a result any info i can bring together will hopefully help me and others to achieve this goal.

As i metioned before - I upgraded my HTPC in the hope that I could significantly reduce it's idle power draw. I figured that apart from perhaps an amd xpress200 chipset/integrated gfx system the 6150 board would give me the lowest power usage with modern day performance. I was perhaps optimistically hoping for sub 60W. When i see in another thread that mike has a similar sytem drawing just 36W idle it begs the question - what in my system is using so much more power than his - and what can i do about it!

Marc

sun.moon
Posts: 174
Joined: Wed Sep 21, 2005 5:57 am
Location: Germany

Post by sun.moon » Wed Apr 26, 2006 3:36 am

so, what is the best, most inexpensive and accurate way to measure power at the wall socket?

ziphnor
Posts: 187
Joined: Mon Dec 22, 2003 2:03 am

Post by ziphnor » Wed Apr 26, 2006 4:33 am

Thomas wrote: Personally, I find it's the efficiency in terms of performance/wattage, there is the interesting issue. As long one have an efficient graphic card, CPU and PSU and the overall wattage is low, then it's good. I dont think picknicking about wattage of RAM blocks and chipsets, will change the overall picture much.
Considering that an efficient CPU in idle low power mode uses 1 - 5W, while the total system power is say 60W id say that nitpicking about the remaining parts is well justified :) The question is of course whether the remaining power draw is pretty much the same for all systems.

cmthomson
Posts: 1266
Joined: Sun Oct 09, 2005 8:35 am
Location: Pleasanton, CA

Post by cmthomson » Wed Apr 26, 2006 5:21 am

sun.moon wrote:so, what is the best, most inexpensive and accurate way to measure power at the wall socket?
I'm happy with the Seasonic PowerAngel: http://www.seasonicusa.com/products.php?lineId=8

Lots of features, easy to use, reusable to measure any appliance over short or long time periods.

It's not cheap though. The cheapest approach (assuming you already own a multimeter, soldering iron and wirecutters) is to buy a $1 extension cord and 5W 0.33 ohm resistor, insert the resistor inline, and measure the voltage drop.

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Wed Apr 26, 2006 10:18 am

cmthomson wrote:
sun.moon wrote:so, what is the best, most inexpensive and accurate way to measure power at the wall socket?
I'm happy with the Seasonic PowerAngel: http://www.seasonicusa.com/products.php?lineId=8
Brennenstuhl do something similar with European pinouts - I have the UK version.
cmthomson wrote:The cheapest approach (assuming you already own a multimeter, soldering iron and wirecutters) is to buy a $1 extension cord and 5W 0.33 ohm resistor, insert the resistor inline, and measure the voltage drop.
Has its limitations - buried somewhere in the archives here is Mike's original article on this. Something to do with AC not being DC?

Back to the topic though. System as per my sig, HTT 254MHz, AC measured at the wall excluding monitor:
59W idle (but HDD spinning)
69W Folding (usual state @ max CnQ)
89W Folding (min CnQ)

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Wed Apr 26, 2006 10:21 am

TheKill-a-watt AC power meter can be found for as little as $20. Every geek and green should have one of these or a Seasonic Power Angel. :wink:

openwheelformula1
Posts: 211
Joined: Thu Sep 22, 2005 9:58 am

Post by openwheelformula1 » Wed Apr 26, 2006 12:16 pm

I meansured mine with kill-a-watt. Less than $20 shipped from buy.com if I remember correctly. I thought everyone here has one already.

dragmor
Posts: 301
Joined: Sun Jul 10, 2005 7:54 pm
Location: Oz

Post by dragmor » Wed Apr 26, 2006 2:06 pm

MikeC wrote:TheKill-a-watt AC power meter can be found for as little as $20. Every geek and green should have one of these or a Seasonic Power Angel. :wink:
Does anyone know where you can get a 220v version. No one seems to stock these in Aus. The only thing that I've seen are $1000 multimeter.

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Wed Apr 26, 2006 10:47 pm

dragmor wrote:Does anyone know where you can get a 220v version.
220V is easy. As before it will take a bit of work to find one for your socket format.

waps
Posts: 15
Joined: Wed Jul 06, 2005 10:39 am
Location: Huizen, Netherlands

Post by waps » Wed Apr 26, 2006 11:21 pm

mesured with energy monitor 3000 (voltcraft/conrad)

idle : 110W
load : 173W (2 x cpuburn)

p4 3GHz, prescot, s478
2 x 512 MB
asus p4p800-e deluxe
nvidia 6800 LE / 256MB
watercooled, eheim 1046
seasonic s12-500
soundblaster audigy2 zs
2x 120mm fans @ 6v
lg dvd multidrive burner
plextor px116a

maarten

sanse
Posts: 399
Joined: Thu Dec 22, 2005 9:48 am
Location: Netherlands

Post by sanse » Thu Apr 27, 2006 12:56 am

asus a8n-sli premium
amd x2 4400
4 gb corsair pc3200
wd raptor 74 gb
samsung spinpoint 250 gb
gigabyte geforce 6600 with ac nv silencer
silentium t2 case with 350 w seasonic psu
creative xifi xtreme
plextor dvd-writer
teac floppy/cardreader
t-balancer fancontroller
7 fans in total

idle 100 watt
load 170 watt (2*cpu-burnin)

Thomas
Posts: 664
Joined: Mon Jun 20, 2005 11:21 pm
Location: Denmark

Post by Thomas » Thu Apr 27, 2006 1:16 am

ziphnor wrote:
Thomas wrote: Personally, I find it's the efficiency in terms of performance/wattage, there is the interesting issue. As long one have an efficient graphic card, CPU and PSU and the overall wattage is low, then it's good. I dont think picknicking about wattage of RAM blocks and chipsets, will change the overall picture much.
Considering that an efficient CPU in idle low power mode uses 1 - 5W, while the total system power is say 60W id say that nitpicking about the remaining parts is well justified :) The question is of course whether the remaining power draw is pretty much the same for all systems.
Ooops, my bad about CPU power - I had my SocketA Sempron in mind :oops:

But I still think it's needed to drill down some more, than measure wall socket power, and list components.

A way to do it, could still be using the wall socket meter, and then disconnect as much as possible; That is USB devices, optical drives, card readers etc.

If components is disconnected one by one, it's possible to measure the power of each component. If more than one RAM block, then measure with one versus two, then you have the power consumption of a RAM block.

At this pont, there will be a much more precise power measuring of the mobo/chipset. In addition, one can substract the VGA idle power, measured by XBit :wink:

It's not relevant for power heat dissation in the case, but in terms of electricity usage, periphals like printers, speakers etc., should also be considered.

Around here, we have USB powered multiconnectors for 220 V. That is, when the USB power shut off, the 220 lines shut off automatically. Thus it's only the PC PSU there still have a stand by usage. Unless it's a 24/7 system, it's effctive and comfortable 8) :D

Image

Linus
Posts: 184
Joined: Fri May 21, 2004 12:47 pm

Post by Linus » Thu Apr 27, 2006 6:30 am

This is a topic I've been interested in forever, but not found many others to share data with.

I can get my PC down to 37W at idle with the following configuration:

Mobile Athlon XP (35W Barton) @ 300MHz
Biostar M7NCG 400Socket A motherboard
2x Kingston 256MB DDR PC3200 CAS3 RAM @ CAS2
Fujitsu MHT2060AH 2.5" 60GB 5400rpm laptop HDD

See my gallery thread: Small & Flexible: Back-of-LCD variable power consumption.

Here's a recent thread about limits on why you don't see systems below ~36W: Laptop vs. Desktop Power Consumption. Basically the consensus is that once you've gone to a truly minimalist system with as many laptop components as are readily available, PSU efficiency at such a low power point and motherboard power draw are the limiting factors.

SPCR articles with interesting wattage data:
Desktop CPU Power Survey, April 2006 (4-5-06)
Power Distribution within six PCs (8-30-05)

Some other threads with wattage measurements:
My Geode PC (1-14-06)
Power consumption info (8-27-04)

ziphnor
Posts: 187
Joined: Mon Dec 22, 2003 2:03 am

Post by ziphnor » Thu Apr 27, 2006 6:58 am

Found out that i could borrow a so-called Spar-o-meter(save-o-meter) at the local library. I dont know how precise it is, but here are my numbers:

System(dedicated HTPC):
A64 3000+ E6
2x256 MB Kingstom DDR400
Asus A8N-E with nb fan removed
Noname 350W PSU modded with 92mm silenx fan
1x92 mm SilenX fan intake fan
Thermal Take Silent Tower (passive, no fan)
Twinhan Cab CI (DVB-C tuner)
ATI X550(passive)
Some Plextor DVD-ROM drive
Samsun SP2504C 250GB SATAII HDD

Tested at the wall socket:

IDLE(winxp desktop)
1Ghz @ 1.1V (CnQ) -> 67-68W
1.8Ghz@ 1.1V (RMClock) -> 68-69W
1.8Ghz@ 1.4V(no PM) -> 73-74W
standby -> 7.5W
off -> 5W

FULL LOAD(CPU Burn-In + RTHDRIBL )
1.8Ghz@ 1.1V (RMClock) -> 85-86W
1.8Ghz@ 1.4V(no PM) -> 100-101W (!)

The interesting thing to note here is that the CnQ underclock had almost no effect( ~1W ), it seems its the undervolt that makes all the difference.
Last edited by ziphnor on Thu Apr 27, 2006 7:30 am, edited 2 times in total.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Thu Apr 27, 2006 7:01 am

For maximum system power draw, CPUBurn by itself is not enough. You should also be running RTHDRIBL concurrently to stress the vidcard. The total AC power draw will go up, even for onboard graphics. BTW, CPUBurn is not the same as CPU Burn-in. The latter does not stress the processor quite as hard; CPUBurn gives slightly higher temperature (and power draw) measurements IRRC, which is why we settled on it as our reference processor stress utility.

QuietOC
Posts: 1407
Joined: Tue Dec 13, 2005 1:08 pm
Location: Michigan
Contact:

Post by QuietOC » Thu Apr 27, 2006 8:08 am

ziphnor wrote: IDLE(winxp desktop)
1Ghz @ 1.1V (CnQ) -> 67-68W
1.8Ghz@ 1.1V (RMClock) -> 68-69W
1.8Ghz@ 1.4V(no PM) -> 73-74W

The interesting thing to note here is that the CnQ underclock had almost no effect( ~1W ), it seems its the undervolt that makes all the difference.
This is confirmation that idle core voltage is the primary power factor.

WinXP does a good job keeping CPUs unloaded by sending halt commands while the system is idle even without C'n'Q. The clockspeed is less relevant if the processor is in a halt state.

C'n'Q will let the processor handle some load while staying in the lowest state. So there is a peak low-state power usage which is not the same as the idle power usage. You can lock the low C'n'Q state (using CrystalCPUID) and test the processor at full load to see the difference between 1GHz @ 1.1V and 1.8GHz @ 1.1V. Then the real question becomes what is cooler a 1GHz CPU at 100% load or a 1.8GHz CPU at 56% load (doing the same amount of work).

Conservation: AMD should lower their default Vcores, and also fix that 1.1V C'n'Q limit--it should be 0.800V at 1GHz. 1.1V could be the new default Vcore for their <2.0GHz processors.
Last edited by QuietOC on Thu Apr 27, 2006 9:36 am, edited 1 time in total.

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Thu Apr 27, 2006 9:17 am

The interesting thing to note here is that the CnQ underclock had almost no effect( ~1W ), it seems its the undervolt that makes all the difference.
We should have a sticky explaining the whys and wherefores about why Vcore (and lowering it) is so important if you want to silence/quiet your computer. Put simply, lowering the Vcore of your CPU to the absolute minimum necessary to ensure reliable POST from cold is the quickest and easiest way I know of reducing the heat (and thus) noise output from a CPU.

The article should include the CMOS power equations etc, because just by looking at the equation you can see why a small change in voltage can lead to a large change in heat output.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Thu Apr 27, 2006 9:52 am

jaganath wrote:We should have a sticky explaining the whys and wherefores about why Vcore (and lowering it) is so important if you want to silence/quiet your computer.
More repetitive wheel reinvention, imo...

CPU Undervolting & Underclocking: A Primer
[url=http://www.silentpcreview.com/article102-page1.html]Undervolting T'Bred-B CPUs with José Ã

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Thu Apr 27, 2006 10:40 am

Fair enough.

Hifriday
Patron of SPCR
Posts: 237
Joined: Thu Aug 05, 2004 3:32 pm

Post by Hifriday » Sat Apr 29, 2006 11:56 am

Measured Apparent Power draw at the wall outlet using a clamp meter (Voltage x Current). However all PSU are external AC/DC bricks and should have PF close enough to 1 that Watts dissipated should be quite close.

==========================
ASUS Pundit P2-AE2 Barebone (Via K8M800)
Sempron 3000+ (Idle 1.1v 0.8Ghz / Load 1.2v 1.8Ghz)
1GB RAM
Samsung Notebook HDD
Slim DVD Burner
Stock case fan plus Zalman 80x15 as CPU fan

IDLE = 31VA : PRIME95 = 52VA : PRIME95+DVD+HDD = 57VA

=========================
ASUS A8NVM-CSM
Opteron 144 TDP 44.1W (idle 0.8v 1.0Ghz, load 1.0v 1.8Ghz)
2x512MB RAM
Hitachi T7K250 250GB HDD
ASUS DVD-ROM
PW200M DC-DC PSU
92mm CPU fan plus 120mm case fan

onboard graphics : IDLE = 47VA : PRIME95 = 65VA
with Geforce 7900GT : PRIME95+3DMark05 = 150VA

==========================
MacMini G4/1.25Ghz
512MB RAM
Samsung 40GB Notebook HDD
Slot optical drive

IDLE = 21VA : Cinebench = 40VA : Sleep = <2VA

MacMini Core Solo/1.5Ghz
512MB RAM
Fujitsu 60GB Notebook HDD
Slot optical drive

IDLE = 25VA Cinebench = 41VA : Sleep = 7VA

=========================
AOpen EZ18 XC Cube SFF barebone (NF2)
Athlon XPM 2200+ CPU @ 1.2v (idle 10x133, load 10x166)
2x512MB RAM
Samsung 40GB Notebook HDD
Samsung Combo DVD drive
PCI TV Tuner
PicoPSU
3 x 80mm fans

IDLE = 52VA PRIME95 = 65VA


Edit: Measured values were Apparent Power (VA) and not True Power (Watts).
Last edited by Hifriday on Mon May 01, 2006 11:10 am, edited 2 times in total.

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Sat Apr 29, 2006 1:05 pm

ASUS Pundit P2-AE2 Barebone (Via K8M800)
Sempron 3000+ (Idle 1.1v 0.8Ghz / Load 1.2v 1.8Ghz)
1GB RAM
Samsung Notebook HDD
Slim DVD Burner
Stock case fan plus Zalman 80x15 as CPU fan

IDLE = 31W : PRIME95 = 52W : PRIME95+DVD+HDD = 57W
Wow! Sub-60W total system power draw at full load! Pretty nifty.

Hifriday
Patron of SPCR
Posts: 237
Joined: Thu Aug 05, 2004 3:32 pm

Post by Hifriday » Mon May 01, 2006 11:03 am

By reading the Voltage x Current with my clamp meter, I was actually only measuring the Apparent Power draw in VA, and not the Real or True Power being dissipated in watts (which should be less due to Power Factor). Thanks to Linus and Devon for setting me straight in this thread.

All the systems I listed above however are using external AC/DC bricks which are switching power supplies and should have Active PFC. This means their PF should be close enough to 1 that although the actual Watts dissipated will be less, it should be within 5% of the VA figure. I've edited the post with the correct units.

In addition I tried loading the Pundit system further using CPUBurn+Rthdribl, however maximum apparent power was still under 51VA. It seems Rthdribl didn't like the K8M800 onboard graphics, it gave me an error at first and asked me to install Direct X SDK, then when it ran it was at 0.04 fps so probably didn't properly load the graphics system. Running just Prime95 or just CPUBurn (K7 high) also resulted in the same 51VA. Running P95 + 3DMark03 + DVD playback + HDD scanning did however manage to produce a higher 57VA.

QuietOC
Posts: 1407
Joined: Tue Dec 13, 2005 1:08 pm
Location: Michigan
Contact:

Post by QuietOC » Mon May 01, 2006 11:25 am

jaganath wrote:
ASUS Pundit P2-AE2 Barebone (Via K8M800)
Sempron 3000+ (Idle 1.1v 0.8Ghz / Load 1.2v 1.8Ghz)
Wow! Sub-60W total system power draw at full load! Pretty nifty.
Hifriday does your 3000+ have the 1.1V minimum voltage lock problem?

Those voltages actually seem a little high for a 90nm Sempron.

Rthdribl requires pixel shaders, and the VIA UniChrome IGP lacks those.

3DMark2000/2001SE would be a better stresser for the UniChrome IGP.
Last edited by QuietOC on Mon May 01, 2006 7:52 pm, edited 1 time in total.

nici
Posts: 3011
Joined: Thu Dec 16, 2004 8:49 am
Location: Suomi Finland Perkele

Post by nici » Mon May 01, 2006 3:14 pm

ASUS A8N-SLI Deluxe
3700+ San Diego
X800GTO
2x512Mb PC3200
Antec Phantom 350 PSU
1x 80Gb 2,5"HDD and 1x250Gb 3,5"HDD
Swiftech MCP350 dc pump.

Think thats about it for now..

I don´t have any optical drives ATM and only one 120mm fan, the pump draws 11W though according to specs.

Idle 115W
FAH 143W
FAH+ ATITool scan for artifacts 185W

Post Reply