The Difference a Few Percent Makes

Hopefully we've made it clear that upgrading an existing power supply to a higher efficiency model purely for the power savings doesn't make sense. However, there are times when you need to buy a new power supply, so we will wrap things up with a closer examination of how efficiency impacts power costs. Should you really care about the difference between 85%, 87%, or 90% efficiency?

This time, we don't need to worry about specific systems, but instead we will focus on efficiency and monetary savings at various power loads. The following table is again a best-case scenario for saving money -- i.e. you are running the system 24/7. Efficiency 1 is the base value and we compare the savings you would gain by selecting a power supply that achieves Efficiency 2. Efficiency ratings at the various loads represent what you might realistically find in various high-end power supplies currently on the market -- so getting 90% efficiency with a load of only 50W isn't going to happen.

Savings from Incrementally Higher Efficiency - 24/7 Yearly Usage
Output - Watts Efficiency 1 Efficiency 2 Savings NC Savings CA Savings GER
50 78% 79% $0.53 $0.91 €1.56
($2.03)
80% $1.05 $1.80 €3.09
($4.01)
81% $1.56 $2.66 €4.58
($5.95)
82% $2.05 $3.51 €6.03
($7.83)
200 80% 81% $2.03 $3.46 €5.95
($7.73)
83% $5.94 $10.13 €17.41
($22.64)
85% $9.66 $16.49 €28.34
($36.84)
87% $13.22 $22.55 €38.77
($50.40)
400 85% 86% $3.60 $6.14 €33.61
($43.70)
87% $7.11 $12.13 €10.55
($13.71)
88% $10.54 $17.99 €20.85
($27.10)
89% $13.90 $23.72 €30.92
($40.19)
700 85% 86% $6.29 $10.74 €18.45
($23.99)
87% $12.44 $21.23 €36.49
($47.43)
88% $18.45 $31.48 €54.11
($70.34)
89% $24.32 $41.50 €71.33
($92.73)

Obviously, the higher the load the better your savings, since a difference of 1W hardly matters. Your best course of action would be to select a power supply that offers the best efficiency at the load you will use most frequently. So for example, if you only play games on your computer and otherwise have it shut off, you might seriously consider a power supply with optimal efficiency at the 500W-600W range. On the other hand, if you typically just surf the Internet you'll probably be more interested in the efficiency at 100W-200W.

At the maximum load of 700W, and going with German power costs, the difference between an 85% and 89% efficiency power supply could be as much as €71. That's enough to get a significantly better power supply, but of course that sort of savings is unrealistic since it will be extremely difficult to achieve a 700W load all the time. The 400W load represents a more realistic maximum, as something like an overclocked quad-core system running Folding@Home could actually draw that much power around the clock. In that case, your savings could still be a pretty significant €30 per year, so over three years you could save almost €100. If you only run the system eight hours per day, however, the difference in cost drops off quickly.

Obviously, spending $20 more just to increase efficiency by 1% isn't necessary. You'll probably use a power supply for at least three years, so all other things being equal higher efficiency is good. That "all other things" is the problem, however, since rarely are the other areas the same. Pay attention to the other features like noise levels, voltage regulation, and the number and type of connector as well. Also keep in mind that we still have changing ATX standards, and sometimes new connectors, so spending a small fortune on a top quality PSU that might be outdated in a year or two might not be the best course of action either.

The bottom line ends up being a simple case of common sense: don't buy more power supply than you actually need, and don't spend a lot of money for a small increase in efficiency. Figure out how much power your system will normally use, and then choose a power supply appropriate for that sort of workload. If you routinely stress your system (i.e. workstation loads or intense gaming), an extra $100 for a high-end power supply might be a good idea. For most users, however, moderation will be the better course of action.

Finally, we spent quite a bit of time putting together the spreadsheet that we used to generate the tables in this article. We selected a few different markets for our power costs, and then we selected several different systems. Obviously, we couldn't cover everything, but for those who are interested in running their own calculations we thought you might appreciate our spreadsheet. Feel free to insert your own KWh costs, efficiency, and system power requirements to see how things change. (The highlighted fields should be the only areas you need to modify.)

Using a Higher Efficiency PSU to Reduce Costs
Comments Locked

59 Comments

View All Comments

  • JarredWalton - Friday, November 14, 2008 - link

    If the laptop isn't plugged in, the power brick should use 0W (or at least less than 1W).
  • MadMan007 - Friday, November 14, 2008 - link

    ...and thi quote is the most important one that made me decide it's not economically meaningful to upgrade from a ~75% PSU to an 85% one. When you do these estimates on non-24/7 use the savings plummet quickly.
  • MadMan007 - Friday, November 14, 2008 - link

    Grr...quote window didn't work right, why can't we just use tags?

    Anyway here's the quote:
    "If you only run the system eight hours per day, however, the difference in cost drops off quickly."
  • Nfarce - Friday, November 14, 2008 - link

    "Hopefully we've made it clear that upgrading an existing power supply to a higher efficiency model purely for the power savings doesn't make sense."

    I am not nor have I ever been concerned how much power my PCs use (or my PS3). Compared to other "hobbies" such as street racing, cruising, spending $50/night bar hopping, and other things people get involved with and in trouble over, PC and console gaming at home is cheap and relatively environmentally friendly. Besides, the logic behind spending hundreds on a higher efficiency PS to lower utility bills is about as brilliant as spending $30,000 on a new hybrid Camry to save money on gas. But if it makes you feel better about yourself, hey, it's *your* money.

    However, as we shift to a new administration in the States next year which has already stated it wants to target the coal industry, I might have a change of tune. We will see utilities skyrocket with the green syndrome of progressing to wind farms and solar power that just won't make up for coal fired plants. We already know the environmentalists and other special interest hacks here will poo-poo on nuclear power.

    Talk to me in two years...
  • Griswold - Friday, November 14, 2008 - link

    About time you share our energy pain in europe, then. :P
    You're still not where we are as far as gasoline goes...
  • 7Enigma - Monday, November 17, 2008 - link

    Then blame your government. Your high gas prices are a direct result of high taxes (likely to pay for the universal healthcare), not that we in the US get a better deal.
  • yyrkoon - Friday, November 14, 2008 - link

    Using less power will *always* benefit a household more than anything else concerning saving money where power is the concern. It is also not just a one item deal when trying to figure out how to cut power costs. Refrigerators/deep freezers commonly in most households use more power than anything else. Microwaves, coffee makers, rice cookers, and hair dryers etc can all use more power, but typically run for far less time. Another place to save on power costs would be changing the type of lightning one uses, say from incandescent lighting to LED lighting.

    However, as has been said by many people before in the past, many many times: there is no such thing as a free lunch. Saving power by using a more efficient light as an example is of course going to cost you more money. But also with LED lights you're going to pay a premium for those more efficient lights. So, in the short term, best way to save money is just to turn that item off when not in use. This goes for VCR's, Computers, or whatever does not need to be plugged *right_now*(and yes, most of us should know that most appliances do draw at least some power when off, but still plugged in). Even going completely off grid(meaning you get your power 100% from solar, wind, or multiple other sources) is going to be just like paying your power bill up front, with reoccurring charges for batteries, and maintenance for your equipment. In case of the latter expect to pay tens of thousands of US dollars just for the price of admission.

    Now, as for as strictly Power Supplies are concerned, Yes a more efficient power supply *will* save you money. How much really depends, and there are other factors to consider than "how efficient it *is*'. You need to determine exactly how much power your system will consume, and procure a PSU that is most efficient at that power level. Just because a power supply is 99.9% efficient does not mean it will work well for your given application. Other factors would be longevity, and reliability. Data centers often purchase PSU's where the given system using them only uses 25-40% of that PSU's capacity. This is why current technology is 'trending' towards power supplies with a better/broader power efficiency range(e.g. they are most efficient on the power curve where they are planned to be loaded at). That said, these types of power supplies used by data centers, etc are not of the off the shelf variety(usually).

  • Staples - Friday, November 14, 2008 - link

    There are a ton of people who leave their computers on 24/7 for no good reason. I am a tree hugger and of course I put mine into S3 sleep if I even walk away for more than 10 minutes. Plus, my second computer is very low power because I bought really low power parts for it including one of the most important, integrated video.

    In my main computer, I have an ATI 4850 which sucks a lot of power even being idle and I have a guilty conscious about even using for non gaming needs. Hybrid VGA power state hardly exists now but I am glad it will be coming eventually because powerful video cards sitting idle is one of the biggest wastes of power. Also, I am glad that Vista has Cool and Quiet built in because most people do not even know you need software to make it work (unlike Intel's speed step which works without any software).
  • cyclo - Saturday, November 15, 2008 - link

    This is where nVidia currently has ATI beat. I'm not sure about nVidia's cards on the lower end of the scale but on the GTX 2xx class of cards, they implement a power saving "2D" mode when the GPU is mostly idling (basically when not playing games or videos).

    On my GTX 260, the GPU core downclocks to 301 (from 621), the shader to 602 (from 1295), and the memory to 200 (from 2052) when I am just surfing the web (which is basically "2D" mode). The clocks go up to default as soon as I start playing a video and of course start playing a game. The temps at "2D" mode goes down to 47 C from 54 C in idle "3D" mode (playing a video).

    There is one problem though and I hope nVidia can fix this with a future driver release. That is when you run 2 monitors the video card never goes into "2D" mode... even when you are not gaming or playing a video. This is why I am forced to disable my 2nd monitor whenever I don't have a need for it.
  • JarredWalton - Saturday, November 15, 2008 - link

    ATI has been doing the same thing for about as long as NVIDIA. There was an issue with 4870 initially where the power saving modes didn't engage properly, but that has been fixed for a while now. NVIDIA is more aggressive, however, on dropping clocks and reducing voltages as well I think.

    Speaking of multi-monitor support, wasn't there a problem with NVIDIA cards and dual monitors with certain 3D engines? Also seem to recall hearing the second display gets shut off in all 3D games on NVIDIA. Maybe that was fixed as well, though.

Log in

Don't have an account? Sign up now