Actual Power Consumption And Current Requirements
Now that it’s clear what the efficiency of a power supply affects, and that all the test measurement values are taken at the socket, you can figure out the actual power consumption by subtracting the losses due to the inefficiency of the power supply. Since the graphics cards have very different load levels, the average used is 82.4% (based on manufacturer statement).
Formula for the calculation:
Actual Consumption in Watts = Measurement at Socket in Watts x efficiency of power supply
300 Watts = 364 Watts x 0.824
This step is necessary because to determine the power supply class, only the actual power consumption of the graphics card is important. The previous value, with efficiency, is only crucial for the calculation of electricity cost.
Unfortunately, now another very important factor comes into play: the current. At these high values, many older power supplies fail when a new high-end graphics card is installed. Not all 500 watts are created equal, especially when it comes to cheap power supplies. Since many components are powered on the 12 V line, the power supply also needs sufficient current (amperage, A).
Let’s take the Radeon HD 4870 as an example. AMD states “500 watt power supply” as a requirement on the packaging. According to the actual measurements, though, the graphics card doesn’t need more than 150 watts even under full load. Now we’ll calculate in 150 watts for the CPU, motherboard and drives, 300 watts should be enough. If we assume from this a 300 watt load on the 12 V line, this means 25 A is required for this rail.
Formula for the calculation:
Current Strength (Amps) = Watt : Volt25 A = 300 Watts : 12 Volts
Cheap power supply 500 watts with 17 A on the 12V line.
As you can see from the stated power rating, you can very quickly run into trouble with this 500-watt power supply, because despite the 500 W figure, it only provides 17 A on the 12 V line—at least 20 A to 25 A is necessary for a current standard PC with the HD 4870. So even if AMD sets 500 watts as a requirement, you can not assume all cheap power supplies will suffice based solely on this parameter.
With current brand name power supplies, things are different: pretty much every 500-watt brand name power supply will provide 25 A on the 12 V line. Even older brand name power supplies with fewer watts achieve the required targets.
Brand name power supply: 433 watts and 33 A on the 12 V line.
If you want to be on the safe side with your power supply, you should determine the maximum power consumption of all components, and then calculate the maximum current needed. Especially when running SLI or CrossFire, the current requirements can quickly become more important than the wattage class with high-end graphics chips (such as the GeForce Ultra 3SLI with 38.8 A at 465.6 watts).
| Actual Power Consumption Nvidia Graphic Cards | 2D Desktop in Watts | 3D Full Load in Watts | Current in Amps (A) |
|---|---|---|---|
| GeForce GTX 280 (1024 MB) | 43.7 | 204.4 | 17.0 |
| GeForce GTX 260 (896 MB) | 38.7 | 191.2 | 15.9 |
| GeForce 9800 GX2 (2x512 MB) | 89.8 | 217.5 | 18.1 |
| GeForce 9800 GTX+ (512 MB) | 45.3 | 126.1 | 10.5 |
| GeForce 9800 GTX (512 MB) | 51.1 | 131.8 | 11.0 |
| GeForce 9600 GT (1024 MB) | 31.3 | 68.4 | 5.7 |
| GeForce 9600 GT (512 MB) | 34.6 | 85.7 | 7.1 |
| GeForce 8800 GTS OC (512 MB) | 51.9 | 142.6 | 11.9 |
| GeForce 8800 GTS (512 MB) | 51.1 | 136.0 | 11.3 |
| GeForce 8800 GT (1024 MB) | 32.1 | 77.5 | 6.5 |
| GeForce 8800 GT (512 MB) | 42.0 | 111.2 | 9.3 |
| GeForce 8800 ULTRA (768 MB) | 74.2 | 172.2 | 14.4 |
| GeForce 8800 GTX (768 MB) | 67.6 | 158.2 | 13.2 |
| GeForce 8800 GTS (640 MB) | 61.0 | 125.2 | 10.4 |
| GeForce 8800 GTS (320 MB) | 51.9 | 112.1 | 9.3 |
| GeForce 8600 GTS (512 MB) | 28.0 | 61.0 | 5.1 |
| GeForce 8600 GTS (256 MB) | 23.9 | 56.0 | 4.7 |
| GeForce 8600 GT (256 MB) | 20.6 | 46.1 | 3.8 |
| GeForce 8500 GT (256 MB) | 17.3 | 29.7 | 2.5 |
| GeForce 8400 GS (256 MB) | 15.7 | 26.4 | 2.2 |
| GeForce 7950 GX2 (2x 512 MB) | 46.1 | 110.4 | 9.2 |
| GeForce 7950 GT (512 MB) | 28.8 | 65.1 | 5.4 |
| GeForce 7900 GTX (512 MB) | 33.8 | 84.0 | 7.0 |
| GeForce 7900 GTO (512 MB) | 33.0 | 80.8 | 6.7 |
| GeForce 7900 GT (256 MB) | 23.9 | 48.6 | 4.1 |
| GeForce 7900 GS (256 MB) | 25.5 | 49.4 | 4.1 |
| GeForce 7800 GTX (512 MB) | 33.8 | 107.9 | 9.0 |
| GeForce 7800 GTX (256 MB) | 33.8 | 85.7 | 7.1 |
| GeForce 7800 GT (256 MB) | 25.5 | 65.1 | 5.4 |
| GeForce 7600 GT (256 MB) | 19.8 | 39.6 | 3.3 |
| GeForce 7600 GS (256 MB) | 16.5 | 27.2 | 2.3 |
| GeForce 7300 GT (256 MB) | 16.5 | 23.9 | 2.0 |
| GeForce 7300 GS (128 MB) | 14.8 | 23.1 | 1.9 |
| GeForce 6800 ULTRA (256 MB) | 47.8 | 80.8 | 6.7 |
| GeForce 6800 GT (256 MB) | 42.8 | 66.7 | 5.6 |
| GeForce 6600 GT (128 MB) | 21.4 | 47.0 | 3.9 |
| Actual Power Consumption Nvidia Graphics Cards (SLI) | 2D Desktop in Watts | 3D Full Load in Watts | Current in Amps (A) |
|---|---|---|---|
| GeForce GTX 280 SLI (1024 MB) | 69.2 | 321.4 | 26.8 |
| GeForce GTX 260 SLI (896 MB) | 75.8 | 379.0 | 31.6 |
| GeForce 9800 GTX SLI (512 MB) | 95.6 | 257.1 | 21.4 |
| GeForce 9600 GT SLI (1024 MB) | 51.9 | 125.2 | 10.4 |
| GeForce 8800 GTS SLI (512 MB) | 91.5 | 243.1 | 20.3 |
| GeForce 8800 GT SLI (1024 MB) | 53.6 | 145.0 | 12.1 |
| GeForce 8800 GT SLI (512 MB) | 69.2 | 199.4 | 16.6 |
| GeForce 8800 ULTRA 3SLI (768 MB) | 221.7 | 465.6 | 38.8 |
| GeForce 8800 ULTRA SLI (768 MB) | 144.2 | 354.3 | 29.5 |
| GeForce 8800 GTS SLI (320 MB) | 94.8 | 222.5 | 18.5 |
| GeForce 8600 GTS SLI (256 MB) | 37.1 | 104.6 | 8.7 |
| GeForce 8600 GT SLI (256 MB) | 29.7 | 84.9 | 7.1 |
| GeForce 7950 GX2 SLI (4x512 MB) | 87.3 | 181.3 | 15.1 |
| GeForce 7950 GT SLI (512 MB) | 42.0 | 110.4 | 9.2 |
| GeForce 7900 GT SLI (256 MB) | 34.6 | 87.3 | 7.3 |
| GeForce 7900 GS SLI (256 MB) | 29.7 | 84.0 | 7.0 |
| GeForce 7800 GTX SLI (256 MB) | 56.0 | 156.6 | 13.0 |
| GeForce 7800 GT SLI (256 MB) | 35.4 | 114.5 | 9.5 |
| GeForce 7600 GT SLI (256 MB) | 21.4 | 67.6 | 5.6 |
| GeForce 7600 GS SLI (256 MB) | 14.8 | 45.3 | 3.8 |
| Actual Power Consumption AMD Graphics Card | 2D Desktop in Watts | 3D Full Load in Watts | Current in Amps (A) |
|---|---|---|---|
| Radeon HD 4870 X2 (2x1024 MB) | 113.7 | 268.6 | 22.4 |
| Radeon HD 4870 (512 MB) | 68.4 | 151.6 | 12.6 |
| Radeon HD 4850 (512 MB) | 47.8 | 109.6 | 9.1 |
| Radeon HD 4670 (512 MB) | 14.0 | 56.9 | 4.7 |
| Radeon HD 4550 (512 MB) | 14.0 | 25.5 | 2.1 |
| Radeon HD 3870 X2 (2x512 MB) | 56.0 | 202.7 | 16.9 |
| Radeon HD 3870 (512 MB) | 25.5 | 92.3 | 7.7 |
| Radeon HD 3850 (256 MB) | 19.8 | 72.5 | 6.0 |
| Radeon HD 3650 (512 MB) | 20.6 | 51.9 | 4.3 |
| Radeon HD 3470 (256 MB) | 21.4 | 29.7 | 2.5 |
| Radeon HD 3450 (256 MB) | 17.3 | 23.1 | 1.9 |
| Radeon HD 2900 XT (512 MB) | 64.3 | 169.7 | 14.1 |
| Radeon HD 2600 XT (512 MB) | 18.1 | 54.4 | 4.5 |
| Radeon HD 2600 XT (256 MB) | 26.4 | 56..0 | 4.7 |
| Radeon HD 2600 Pro (256 MB) | 14.8 | 38.7 | 3.2 |
| Radeon HD 2400 XT (256 MB) | 14.8 | 28.0 | 2.3 |
| Radeon HD 2400 Pro (256 MB) | 14.0 | 21.4 | 1.8 |
| Radeon X1950 XTX (512 MB) | 31.3 | 110.4 | 9.2 |
| Radeon X1950 XT (256 MB) | 33.8 | 123.6 | 10.3 |
| Radeon X1950 Pro (256 MB) | 31.3 | 77.5 | 6.5 |
| Radeon X1950 GT (256 MB) | 29.7 | 69.2 | 5.8 |
| Radeon X1900 XT (512 MB) | 35.4 | 130.2 | 10.8 |
| Radeon X1900 XT (256 MB) | 30.5 | 106.3 | 8.9 |
| Radeon X1900 GT (256 MB) | 34.6 | 83.2 | 6.9 |
| Radeon X1800 XT (512 MB) | 39.6 | 88.2 | 7.3 |
| Radeon X1800 XT (256 MB) | 37.1 | 91.5 | 7.6 |
| Radeon X1800 XL (256 MB) | 33.8 | 57.7 | 4.8 |
| Radeon X1800 GTO (256 MB) | 29.7 | 48.6 | 4.1 |
| Radeon X1650 XT (256 MB) | 29.7 | 58.5 | 4.9 |
| Radeon X1650 Pro (256 MB) | 23.1 | 44.5 | 3.7 |
| Radeon X1600 XT (256 MB) | 23.1 | 42.8 | 3.6 |
| Radeon X1600 Pro (256 MB) | 23.9 | 38.7 | 3.2 |
| Radeon X1300 XT (256 MB) | 24.7 | 44.5 | 3.7 |
| Radeon X1300 Pro (128 MB) | 19.8 | 29.7 | 2.5 |
| Radeon X1300 (256 MB) | 18.1 | 26.4 | 2.2 |
| Radeon X800 XT (256 MB) | 23.9 | 64.3 | 5.4 |
| Actual Power Consumption AMD Graphics Card (CrossFire) | 2D Desktop in Watts | 3D Full Load in Watts | Current in Amps (A) |
|---|---|---|---|
| Radeon HD 4870 CF (512 MB) | 146.7 | 293.3 | 24.4 |
| Radeon HD 4850 CF (512 MB) | 93.1 | 216.7 | 18.1 |
| Radeon HD 3870 X2 CF (4x512 MB) | 111.2 | 416.9 | 34.7 |
| Radeon HD 3870 3CF (3x512 MB) | 80.8 | 296.6 | 24.7 |
| Radeon HD 3870 CF (512 MB) | 49.4 | 180.5 | 15.0 |
| Radeon HD 3850 CF (256 MB) | 38.7 | 144.2 | 12.0 |
| Radeon HD 3650 CF (512 MB) | 39.6 | 107.9 | 9.0 |
| Radeon HD 2900 XT CF (512 MB) | 130.2 | 359.3 | 29.9 |
| Radeon HD 2600 XT CF (512 MB) | 33.8 | 117.0 | 9.8 |
| Radeon HD 2600 Pro (256 MB) | 28.0 | 77.5 | 6.5 |
| Radeon X1950 XTX CF (512 MB) | 67.6 | 225.8 | 18.8 |
| Radeon X1900 XT CF (512 MB) | 70.0 | 247.2 | 20.6 |
| Radeon X1600 XT CF (256 MB) | 47.8 | 86.5 | 7.2 |
| Radeon X1600 Pro CF (256 MB) | 44.5 | 73.3 | 6.1 |
- 1 / 2
- Next
-
Now, how about £ for the UK site, and € for Europe?
Other than that, a good article. Very informative. Very linkable!
Out of interest did you test these cards with one monitor connected or two?
I mention this because using a power meter I noticed a strange 'bug' on the Nvidia GTX 280. With 2 monitors connected instead of 1 the idle power consumption of the whole PC increased by 38w!!
I tested this with multiple driver revisions and the latest available BIOS. Turns out I'm not the only one to see this either.
I don't get this problem with my 8800GT 512mb. The power consumption of that is the same with one or two monitors.
Also out of interest why does the 8800 1024Mb use less power than the 512mb model, is it a 55nm core?
Nice you tested all those cards
The cost is not that relevent really because using different quality psu's at differing variations of optimal load will produce wildly varying results not to mention the fact that various utility suppliers charge different rates. Therefore, I see no real need for a conversion to pounds or euro's for those of us living in areas outside of the US. What is important and interesting about this article is that it allows people to make an informed choice about what psu is likely to be best for their chosen setup by using the figures in a comparative way regardless of cost. Great article.
small reminder for "green" people.
ask yourself following questions:
Does my house use electricity/oil for heating?
if yes how much?
conclusion: your computer reduces heating costs the same amout it consumes energy 1:1.
ofcourse same applies to lights/tv/etc
when calculating "waste" energy you really need to dig deep before getting correct answer.
well that argument isn't entirely correct because certain things like lights don't radiate heat into the room it stays near the ceiling.
also high temperatures reduce the lifespan of a pc - and require more airflow and cooling which will relate to more dust and the need for more regular maintainanceof the PC.
also during the summer time when it is hot you don't require heating so any heat generated then IS wasted energy.
generally speaking wherever any energy is wasted it is better to eliminate it because 9 times out of 10 it will pay off.
Every single article I have read today has someone complaining about dollars and pounds.
It would take less time to lookup a local price, do a conversion or order the thing from the US.
whining pomm stereotype is never going to go away.
This "mi1ez " guy has no life eithere that or his from a different site and doesn't like this one.. Why don't you go to every site on the net that's in dollars and tell them change to GBP just because your to lazy to do a conversion.. Get a life
er fookdis....I don't know what country you live in but here in UK it's kinda hard to "order the thing from the US" when we are talking about power providers. As for conversions its usually pretty easy since the pound is generally worth a little under 2 US dollars. But I cartainly agree that to niggle about the fact that the article only gives prices in US currency for the reasons I already stated above. I'm sure the author of the review only included a cost comparison as a rough guide anyway. On the other hand, there is an underlying issue here in which milez actually has a point. Every article I've ever read here is exactly the same whether you read it on the "uk content" or "us content" pages and sometimes this causes incorrect comparisons as you can see if you read my response to the recent 3x2mb ddr3 memory test.
"...since the pound is generally worth a little under 2 US dollars"
Those were the days. Happy, recession free days.
Fair point. It's more like 1.65 atm. But the other problem with buying "cheaper" gear from the US is that by the time you add the postage it's actually MORE expensive than the inflated UK prices. For example, the postage charge on motherboards from US to UK is around the $70+ mark. The point I'm making (off topic) though is that when reviews start doing hardware test conclusions based on an American concept of "value" there is definately a "lost in translation" problem and it is THIS that should be looked at when considering content changes between reviews on US and UK Content pages.
1.65?!
It is about £1 - $1.38 at the moment. Which is changing daily. You really wouldn't want to be buying anything from either the US or Eurozone at the moment.
Ha Ha,
Maybe they should evacuate everyone from the UK across the channel in little fishing boats and let the island sink under the weight of the debt it's carrying. No debt - no problem. (Except for the french having to put up with the englishers.)
Dubbya. We already did that before...we sent all the criminals to America and Australia (and most of those who went on their own were fleeing the threat of the debters prisons). But let's not get into some petty one upmanship that you don't have the mental capacity to persue eh.
Out of interest did you test these cards with one monitor connected or two?I mention this because using a power meter I noticed a strange 'bug' on the Nvidia GTX 280. With 2 monitors connected instead of 1 the idle power consumption of the whole PC increased by 38w!!I tested this with multiple driver revisions and the latest available BIOS. Turns out I'm not the only one to see this either.I don't get this problem with my 8800GT 512mb. The power consumption of that is the same with one or two monitors.Also out of interest why does the 8800 1024Mb use less power than the 512mb model, is it a 55nm core?
Indeed it seems strange that the 1024 Mb configuration would consume more power, is that perhaps a mistake on the table or test?
On a review on www.techpowerup.com about the VVIKOO GeForce 8800 GT, it is written: "In Idle we see no difference between the 1024 MB and the 512 MB version of the 8800 GT. Under load this changes slightly, with the 1 GB version taking a small 10W lead (~5%). The peak power draw is higher as well with 231 W vs. 213 W on the 8800 GT 512 MB. This increase in power draw was to be expected since the additional eight memory chips will consume some power."
To all the posters complanting about us$ used
"We used 20 cents (the price at the Munich utility company Stadtwerke München) for the electricity calculation. PSU stands for the power supply unit." so they used euro's which are 1:1 with the pound quote taken from page 2
Very good article, I had been thinking for years that the power consumption for the graphics cards is going maddly off the scale. I have an 2nd PC used by the family thats has an old TI4600 card and I would like to upgrade it because some DirectX 9 games fail to work but its only a 250watt Shuttle case+psu. Anyone any ideas how much a TI4600 draws and what I could swap it for without killing the shuttle psu?
Being one that is currently experiencing power issues I found this to be information I desperately needed!
So will the power consumption start playing into the monthly value shakedowns? Speaking of which, where's the "Best Video Cards For The Money: Jan '09" article?