Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
Power Hardware

Is It Worth Investing In a High-Efficiency Power Supply? 328

MrSeb writes "If you've gone shopping for a power supply any time over the last few years, you've probably noticed the explosive proliferation of various 80 Plus ratings. As initially conceived, an 80 Plus certification was a way for PSU manufacturers to validate that their power supply units were at least 80% efficient at 25%, 50%, 75%, and 100% of full load. In the pre-80 Plus days, PSU prices normally clustered around a given wattage output. The advent of the various 80 Plus levels has created a second variable that can have a significant impact on unit price. This leads us to three important questions: How much power can you save by moving to a higher-efficiency supply, what's the premium of doing so, and how long does it take to make back your initial investment?"
This discussion has been archived. No new comments can be posted.

Is It Worth Investing In a High-Efficiency Power Supply?

Comments Filter:
  • The Maths (Score:5, Informative)

    by Press2ToContinue ( 2424598 ) * on Thursday December 13, 2012 @08:00PM (#42282227)

    new efficiency @ load % - old efficiency @ load % = delta%
    integrate over time (delta%*cost kw/hr) until result = new unit cost (solve for t)

  • Cooler (Score:2, Informative)

    by Anonymous Coward on Thursday December 13, 2012 @08:08PM (#42282307)

    One advantage of a more efficient PSU is that it runs cooler. This is nice at least if you are going for a silent system, as less fans are then required.

  • Re:The Maths (Score:5, Informative)

    by Carnildo ( 712617 ) on Thursday December 13, 2012 @08:21PM (#42282461) Homepage Journal

    Or don't: it comes out at several tens of years in any realistic scenario.

    Scenario 1: an always-on computer running near-idle for four years.

    Idle power draw, 85% efficient PSU: 66 watts
    Idle power draw, 80% efficient PSU: 70 watts
    Delta: 4 watts
    Total power difference over the four-year life of the computer: 140 kilowatt-hours.
    At 5.5 cents per kilowatt-hour (cheapest power in the US), building with a more-efficient power supply makes sense if it costs no more than $7.70 beyond what the less-efficient power supply does.

    Scenario 2: an always-on computer running Folding@Home for four years using both CPU and GPU.

    Power draw, 90% efficient PSU: 215 watts
    Power draw, 80% efficient PSU: 245 watts
    Delta: 30 watts
    Total power difference over the four-year life of the computer: 1.05 megawatt-hours.
    At 36 cents per kilowatt-hour (most expensive power in the US), building with a more-efficient power supply makes sense if it costs no more than $378 beyond what a less-efficient power supply does.

    The second scenario represents someone running F@H on a modern high-end computer in Hawaii -- not exactly "unrealistic".

  • by anyaristow ( 1448609 ) on Thursday December 13, 2012 @08:30PM (#42282545)

    A PSU has a power efficiency curve that looks like this [anandtech.com]. That article also explains what I'm about to summarize:

    Pick a PSU that is no more powerful than you need, to keep your system in the middle of that curve, for maximum efficiency. 100% margin is more than plenty, so if your components will use 250W max, you don't need a 900W PSU. Look for something in the 500 range, or even less if you pick a good-quality PSU.

    You probably won't be able to make a cost argument for maximizing efficiency, but you can build a quieter system focusing on efficiency, and it's quite satisfying obsessing over something different.

  • by Anonymous Coward on Thursday December 13, 2012 @08:32PM (#42282585)

    This is why "software engineer" is a term I will never use willingly. It is an insult to real engineers. Heat pumps do in fact put more heat into their hot side than they consume in work. They take heat from a low temperature resivoir and send it to a high temperature resivoir.

  • Re:The Maths (Score:2, Informative)

    by jamesh ( 87723 ) on Thursday December 13, 2012 @08:40PM (#42282663)

    new efficiency @ load % - old efficiency @ load % = delta%
    integrate over time (delta%*cost kw/hr) until result = new unit cost (solve for t)

    You're missing the savings on removing that excess heat from your house too (in climates where that is relevant).

    In a cold climate where you are heating your house, unless you can get better $/unit heating out of something else, the "waste" energy is heating the house anyway so it doesn't matter much.

    In a hot climate where you are cooling your house, every unit of heat that you put into the house has to be removed. Firstly from the computer by making the fans work harder, then from the house itself by making your AC work harder.

    Please revise your maths accordingly.

  • Re:More maths (Score:5, Informative)

    by AdamWill ( 604569 ) on Thursday December 13, 2012 @09:03PM (#42282849) Homepage

    That was true in the past when the PSU wasn't a particularly valued component and the industry standard method of rating their power output was 'think of a number, any number. Now write that number on the side.'

    It's *less* true these days if you're buying from one of the decent brands. The numbers they write on their spec sheets actually bear some kind of resemblance to reality, these days: you can actually accurately spec up your expected draw against the capabilities of a PSU and expect it to more or less work out. It's worth leaving a bit of safety room, but you don't really need 2X.

  • Re:The Maths (Score:5, Informative)

    by hawguy ( 1600213 ) on Thursday December 13, 2012 @09:06PM (#42282887)

    Not to mention reduced heat output (and potentially less fan noise due to lower heat), important in many scenarios

    Plus you have to add in costs due to the extra air conditioning load in the summer time (gotta remove all of that heat), and subtract in the winter time to account for the fact that your furnace needs to do less work to keep your house warm.

  • by MtHuurne ( 602934 ) on Thursday December 13, 2012 @09:43PM (#42283177) Homepage

    If you look at efficiency graphs, you'll see that power supplies are typically the most efficient under moderate load: at low and high load the efficiency drops. A typical desktop or home server is idle most of the time, so idle efficiency will have a big impact on the total efficiency. If you over-dimension your power supply, your idle load might be 10% or less of the max rating, which is far from the optimum of the efficiency curve.

    I'd recommend getting a power supply that can deliver a bit more than what you need, for example 450 W if you think you need 350 W max. A bit of margin is useful since you might not have found the actual worst case or you might want to add components later. Also it avoids poor efficiency at the high side of the curve when the system is under load.

  • by sdguero ( 1112795 ) on Thursday December 13, 2012 @10:28PM (#42283453)
    Disagree about peak efficiency. In my experience testing PSUs, it is normally found around 90% load. Newer PSUs have gotten a lot better and enhancing efficiency at lower load levels, but PSUs still work most efficiently when running near the load they are designed for.

    Newegg's calculator is a joke. It drastically overestimates requirements so they can pimp massive PSUs with higher profit margins. I suggest adding up the various component manufacturer specifications (i.e. max power draw of the MB, GPU(s), HDD(s), DIMM(s), and CPU(s)) and throw in 10-15 W for overhead, then buy a decent PSU with a load rated as close to that number as you can get. Even with a dual GPU setup, you are VERY unlikely to exceed 400W of DC power draw. My current mid-range single GPU system draws around 200W under load (gaming).
  • by OneAhead ( 1495535 ) on Thursday December 13, 2012 @10:41PM (#42283525)

    So they work well in heating a house as long as its not cold outside. Probably not so good in a real winter..

    Theoretically spoken, they just need to pump against a larger gradient if it's really cold. They will still have a benefit, only less.

    Now, practically spoken, there are these nasty little engineering considerations. A practical heat pump has to be built for cold climates, and the heat pump/AC combos that are popular in the warmer parts in the US aren't, and are actually capable of being slightly less efficient than a resistor if it's really really cold outside.

  • by OneAhead ( 1495535 ) on Thursday December 13, 2012 @10:59PM (#42283615)
    While GP is woefully incorrect and you're right to call him out on it, your explanation isn't right either. Heat pumps can in fact pump against a gradient, and are mostly used to pump heat from a cold to a hot place. Air-source heat pumps (ie. coupled to the outside air rather than a geothermal reservoir) are used in parts of the US to heat houses in the winter and cool them in the summer. They're also what makes a refrigerator work. A fridge pulls heat from a cold place (inside the fridge) to a warmer place (outside the fridge). The resulting decrease in entropy needs to be balanced by an equal of greater increase in entropy, which is accomplished by converting electricity to heat. Or, to avoid the thermodynamic jargon, you're pumping against a gradient, so you need to spend energy to do so. The heat produced at the back of your fridge is the sum of the heat that was pulled out of the interior of the fridge + the heat-equivalent of the electricity the fridge consumed. This is also what an A/C does. Now, if we turn the A/C inside-out, so that it pumps heat from outside to inside, then you have the kind of heat pump we use to heat our homes in the winter. The sum of the heat that was pulled from outside and the heat-equivalent of the electricity the device consumes is larger than the heat-equivalent of the electricity alone, thus the pump brings more heat into your home than a resistor using the same amount of electricity. GP suggested to generate electricity from this heat gradient, but the flaw in his thinking is that the heat pump as well as any electricity generation device he can come up with are bound by the Carnot efficiency, so you can never get more electricity out than you put in.
  • Re:More maths (Score:5, Informative)

    by Anonymous Coward on Friday December 14, 2012 @01:14AM (#42284289)

    You're thinking of hardwaresecrets.com - they do the type of PS reviews only an EE truly appreciates! :)

  • Re:The Maths (Score:4, Informative)

    by beelsebob ( 529313 ) on Friday December 14, 2012 @07:22AM (#42285497)

    600W is actually pretty enormous by modern standards. an i7 3770k will use 77W, a very high end GPU 250W, and motherboard and other bits about 50... so even for very very high end systems, you're talking about 400W total consumption with everything under maximum load. Under most normal usage you're talking more like 100-200W.

  • Re:The Maths (Score:2, Informative)

    by Man Eating Duck ( 534479 ) on Friday December 14, 2012 @09:55AM (#42286217)

    It doesn't stop the uneducated from buying 1000W power supplies.

    I have a quad core i7 3.2ghz powerhouse with 2high end video cards and 4 drives in it and it's happy with the 400 watt power supply I put in it even at full load. Some people think bigger numbers are better, Others actually calculate what they need.

    A while ago I built my new system using an el-cheapo 400W PSU I had laying around. The system was flaky under load, even though my components couldn't have drawn more than about 250W. I suspect voltage fluctuations even if it might have been able to supply the full 400W. I bought an upper-mid-range 600W PSU to replace it, which I have been very happy with it through the next two upgrades (which makes it about five years), I'm still using it in my gaming rig.

    It might be overkill, but in the long run it was actually a very cheap investment.

  • Re:More maths (Score:4, Informative)

    by fgouget ( 925644 ) on Friday December 14, 2012 @12:18PM (#42287667)

    That brings home another benefit of picking a high efficiency power supply: generally a much higher quality and specs that you can actually trust. For instance compare the review of the Coolmax 750W [hardwaresecrets.com] with that of the Corsair VX450W [hardwaresecrets.com]. The el-cheapo 750W PSU blew up twice [hardwaresecrets.com] after they pulled just 500W while the 450W one managed to provide a stable 572W [hardwaresecrets.com] before it shutdown cleanly due to over load protection! So before buying a power supply it's worth reading a proper review of it, even if you only read the conclusion [hardwaresecrets.com] page [hardwaresecrets.com].

    So just looking at much is saved on electricity is missing the big picture.

Don't get suckered in by the comments -- they can be terribly misleading. Debug only code. -- Dave Storer