I’ve been toying around with updating my computer lately, and one of the topics I wanted to look at was the choice of power supply. For the most part, we’ve long since moved beyond the days where power supplies that cost under $60 are garbage. There are plenty of decent power supplies available, particularly if you don’t mind taking a step down from the latest and greatest in terms of efficiency. Anyway, I was helping a friend put together a new PC the other day and it got me curious.

First, let’s start with the system build he put together, with some input from me. Note that many of the parts were selected based on price and availability on Amazon.com, as that’s where he wanted to purchase the parts (with his Amazon Prime account). In some cases, prices have changed since the purchase a week ago, so shop around as needed. Also note that he used a GTX 780, but I’ve also run some power use tests with just the iGPU as well as with a low-end Radeon HD 7750.

Custom-Built Intel Haswell PC
Component Description Price
Processor Intel Core i5-4670K
(Quad-core 3.4-3.8GHz, 6MB L3, 22nm, 84W)
Motherboard ASRock Z87M PRO4 (mATX) $135
Memory Corsair Vengeance 2x8GB DDR3-1866
(9-10-9-27, CMY16GX3M2A1866C9)
NVIDIA Graphics EVGA GeForce GTX 780 3GB
(2304 CUDA Cores at 967-1020MHz, 6GHz GDDR5)
AMD Graphics Sapphire Radeon HD 7750 1GB (Alternative GPU)
(512 Cores at 800MHz, 4.5GHz GDDR5)
Integrated Graphics Intel HD Graphics 4600
(20 EUs at 350-1200MHz)
SSD Corsair Neutron 256GB $187
HDD Western Digital 2TB Mainstream (WDBH2D0020HNC) $80
Optical Drive ASUS 24X DVDRW SATA (DRW-24B1ST) $22
Case Silverstone PS07B (mATX) $79
Power Supply Antec EA-450 Platinum (450W) $70
Power Supply Cooler Master GX-450 (450W Bronze) (Alternative) $49
Operating System Windows 7 Home Premium 64-bit (OEM) $89
Total Price (not including tax or shipping, with GTX 780 and Platinum PSU) $1592

Now I’m not equipped to tell you about the quality of voltage regulation, ripple, or anything like that, but I happened to have a 450W 80 Plus Bronze PSU that I could use as a comparison point, so I asked if I could take some quick measurements once the system was put together. He agreed, and I ran through a few typical scenarios, summarized in the table below. (Note that I had to use a Molex to 8-pin PEG power adapter in order to run the GTX 780 on the old Thermaltake PSU; everything worked, but that wouldn’t be a solution I’d be comfortable with long-term.)

Bronze vs. Platinum System Power Draw (Kill-A-Watt)
Test Load Thermaltake
Litepower 450W
Antec EA-450
EarthWatts Platinum
24/7 Yearly Savings
iGPU – Idle 34 27 61.4 kWh (~$7.68)
iGPU – Cinebench Single-Core 63 55 70.1 kWh (~$8.76)
iGPU – Cinebench Multi-Core 96 86 87.7 kWh (~$10.96)
iGPU – Cinebench OpenGL 103 92 96.4 kWh (~$12.05)
AMD 7750 – Idle 45 38 61.4 kWh (~$7.68)
AMD 7750 – 3DMark 131 121 87.7 kWh (~$10.96)
NVIDIA 780 – Idle 48 41 61.4 kWh (~$7.68)
NVIDIA 780 – 3DMark 348 325 201.6 kWh (~$25.20)

For power costs, we’re looking at the worst-case scenario of leaving a system on 24/7, which really isn’t realistic unless you’re talking about a server. For a typical PC that’s on eight hours a day, using the US national average price for electricity ($0.125 per kWh), we’re looking at electrical savings of anywhere from $2.56 to $8.40 per year. That may not seem like much, but considering any decent power supply should last five years and you’re looking at $12.80 to $42 in savings. That’s for a $21 difference in upfront costs, which is much smaller than what we’ve seen in the past for the most efficient power supplies – and note that the price difference tends to grow substantially when you’re shopping for 800-1200W PSUs, though that’s perhaps a topic for another day.

For a system that mostly sits idle, you won’t quite break even by going with an 80 Plus Platinum power supply. However, if there’s a graphics card installed and you do a moderate amount of gaming you should eventually come out ahead. More extreme use cases (e.g. 24/7 Folding@Home) start to rapidly recover the initial investment in a quality power supply, and when you consider the reduced heat and noise that comes with having a more efficient PSU, it’s definitely a worthwhile upgrade. Not all 80 Plus Platinum power supplies are created equal, of course, but generally speaking the electronics and engineering required to get that certification also come with a higher level of quality than what you’ll find in lower efficiency PSUs.

As far as the above system build is concerned, I like most of the component selections, but I’m not totally sold on the case. The Silverstone PS07B looks nice enough, but getting all of the wires connected can be a bit difficult at times. The SSD location on the bottom in particular is a bit of a pain, and the power supply location at the top has some silliness to go with it. The PSU location appears designed to work with the PSU upside down (i.e. fan facing upwards and drawing in fresh air), but the manual for the case notes that if you have a PSU with a 120mm or larger fan, they recommend installing it with the fan facing down. I'm not sure that I've ever seen a PSU with a fan smaller than 120mm where you have ventilation through the bottom, but whatever. If you follow the manual’s instructions, that means the ventilation holes in the top of the case along with the magnetic dust filter are completely pointless. I ignored the instructions and installed the power supply in the most sensible manner for the case, but my advice is to look at some of the other good mATX cases.

The Silverstone PS07B isn’t a bad case, but it’s not perfect either, and for the price I think you can do better. (Apologies to my friend for his taste in cases….) For mATX, I’d at least give the BitFenix Prodigy M or Corsair’s Obsidian 350D a look. If you want a larger case, there are tons of options to sort through, depending on what you're after. Other changes you might consider include sticking with 8GB RAM (2x4GB) initially, going with a midrange GPU like a GeForce GTX 770 or Radeon R9 280X, and there are quite a few motherboard options to consider as well. The i5-4670K still strikes a nice balance between price and performance, and with a bit of overclocking you can stretch its legs a bit further.

Anyway, that’s my little two-for-one special for the day: a quick look at the difference in power use you can expect from 80 Plus Bronze vs. Platinum (obviously 80 Plus Silver and Gold will be closer in power use, but they’re also closer in price), along with a list of parts that I’ve recently used in a friend’s PC. If you have any recommendations or complaints with the build, sound off in the comments.

Comments Locked


View All Comments

  • Flunk - Thursday, October 31, 2013 - link

    The only issue with using 80 plus ratings is that they're only "at least" ratings, and they don't actually test the units, even if they do test one unit (and they don't even always do that) it's always one hand-picked unit from the manufacturer and some manufacturers have been caught providing better hardware for testing than they actually sell to the channel. This is normally ok if you're buying from reputable suppliers but if you buy from someone like Thermaltake that has a rather checkered history (or worse a one with a horrible reputation like Diablotek) you never know what you're getting without major research in advance.

    Not only that, many models of power supply vastly exceed their 80 plus efficiency ratings. You have to be pretty careful when it comes to 80 plus ratings because the're not always accurate.
  • ShieTar - Thursday, October 31, 2013 - link

    Not all reviews work like that. German c't magazine is anonymously buying their review samples from shops. Of course they have 30 million € per year of revenue, so they can operate different than most other reviewers.
  • ruzveh - Thursday, October 31, 2013 - link

    Its not entirely the right theory to measure. You have missed onto measure the usage of the power while measuring the units. How much power was utilized in each case while running the PSU? In this case if CPU is powered in idle mode and the PSU is powered only 10% then the efficiency will reduce in both the case.

    Suppose if i buy 300W power supply unit which has bronze certification and another 1000W with Platinum certification then definately anyday anytime your bronze will give u more savings in ur bills as u wont be running heavy apps and games 24x7.. Which is the case with me.. i leave my pc 24x7 for surfing and streaming... i hardly game or run heavy apps..

    Got it?
  • Death666Angel - Thursday, October 31, 2013 - link

    Uhhh, lol. He is comparing the same system, same loads, with 2 different 450W PSUs. Of course his findings are only applicable to a system similar like his. He only says that all else being equal (that includes wattage of the PSU), the platinum one will save you electricity.
  • JarredWalton - Thursday, October 31, 2013 - link

    This is why I used the same system running the same loads with the same size PSU. I leave my system on 24/7 as well, simply because I like all my email to be downloaded and ready when I come to my PC, plus I frequently need to access it remotely (via TeamViewer).
  • fic2 - Thursday, October 31, 2013 - link

    That is why I didn't understand the statement:
    "For power costs, we’re looking at the worst-case scenario of leaving a system on 24/7, which really isn’t realistic unless you’re talking about a server. "
    Most people I know leave their PCs on 24/7. Not to mention most company desktops are on 24/7. Just using that the more efficient PS will pay for itself in 3 years.
  • erple2 - Thursday, October 31, 2013 - link

    Wow. No - my company (and any company that keeps track of their wasted expenses) aggressively encourages sleeping your computer when you leave for the night. My building alone has about 300 employees, so using a machine (even at idle) for 8 hours a day (or more realistically 12 hours a day - I'm an American), and sleeping when you leave, that still is a reasonably significant couple thousand dollars a year savings in electricity alone. The nice thing is that since the company has started to aggressively promote that, they're also aggressively tracking how much power is used. At the end of the fiscal year, they spend 1/2 of the savings on a building party. OK, it's not super-awesome, but 1/2 of about 4000 dollars is still some decent cookout food.
  • Pinkynator - Thursday, October 31, 2013 - link

    I'm too lazy to calculate the average European electricity price, but I'll estimate it at around 0.18€/kWh, which equals $0.25/kWh, or double the price used in this article.

    It's still not a huge saving per year, but taking into account that electricity prices will go up, and that a platinum PSU will be used for upwards of 5 years, things aren't that bad as they seem, even if you only look at the costs of electricity. The feeling of safety, knowing that you're using the best, is priceless.
  • slate_dk - Thursday, October 31, 2013 - link

    In Denmark it is $0,40/kWh with todays exchange rate, so you can tripple the savings.
    Side note: we get a little bit better efficiency as we are running 240 volt.

    So here it pays to save on energy consumption, which is the original motivation behind the energy taxes back in 1972. And it works, we use less than half the energy pr capita compared to the US.
  • ShieTar - Thursday, October 31, 2013 - link

    Overall energy consumption is one thing, but there is an impact of quality of life included in that mp. What I find even more interesting is energy use normalized by GDP:


Log in

Don't have an account? Sign up now