Actual System Power Costs

On the previous page, we estimated the efficiency for each system as being 82%. Obviously, that's not a true way of calculating power requirements or efficiency. Now we're going to shift to the real world and see what the three sample systems end up costing on an hourly basis.

Before we get to the tables, it's important to remember that just because a power supply advertises 90% efficiency doesn't mean you'll always reach that level. You can look at any of our power supply reviews -- or just read Debunking Power Supply Myths -- to understand this better. The short summary is that all power supplies have an efficiency curve, which depends on the load you place on the power supply.

At lower loads and maximum load, efficiency is lower than if you run at a medium load (relative to the PSU's rated output). If you're just surfing the Internet, writing a document, or viewing pictures your system will largely sit idle. Playing a game, doing 3D rendering, encoding a video, or other complex calculations will place a higher load on your PSU. The following tables use actual efficiency with a real power supply to calculate power costs.

System 1

Our entry-level system, System 1, will utilize the Thermaltake TR2 QFan 300W power supply we recommended in our last article. System 1 consumes 90W to 140W of power, depending on load -- those are best-case/worst-case figures. We haven't posted our review of the QFan yet, but it achieves 82% efficiency at 90W load and 84% efficiency at 140W load. The hourly power costs are:

 System 1 Power Costs with Real Efficiency Load Efficiency Outlet Power Cost/hr NC Cost/hr CA Cost/hr GER 90W 82% 110 \$0.008 \$0.014 €0.024 (\$0.031) 140W 84% 167 \$0.013 \$0.021 €0.037 (\$0.048)

System 2

System 2, our midrange system, will use the OCZ ModXStream Pro. This system requires between 160W and 350W of power. The OCZ power supply runs at 84% efficiency for 160W and 85% efficiency for 350W. That gives the following power costs:

 System 2 Power Costs with Real Efficiency Load Efficiency Outlet Power Cost/hr NC Cost/hr CA Cost/hr GER 160W 84% 190 \$0.014 \$0.024 €0.042 (\$0.054) 350W 85% 412 \$0.031 \$0.053 €0.091 (\$0.118)

System 3

Lastly, our high-end system is running two graphics cards for maximum performance. This time we selected the OCZ EliteXStream 800W PSU. Note that even this beefy system still only requires 550W at maximum load, whereas it idles at 310W. In this case, efficiency is 84% idle and 83% at full load.

 System 3 Power Costs with Real Efficiency Load Efficiency Outlet Power Cost/hr NC Cost/hr CA Cost/hr GER 310W 84% 369 \$0.028 \$0.047 €0.081 (\$0.106) 550W 83% 663 \$0.050 \$0.085 €0.146 (\$0.190)
Calculating Power Requirements and Costs Using a Higher Efficiency PSU to Reduce Costs

• #### walk2k - Tuesday, November 18, 2008 - link

If you're curious to find out what your system actually uses, for about you can buy a P3 "Kill-a-Watt" power meter from Amazon for about \$24. This is a doo-hicky that plugs into the power outlet, and then you plug whatever you want to measure into that. I've personally found it very enlightening.

My current PC uses about 120 watts idle in XP, up to about 150 watts when actually doing stuff (cpu slightly loaded, HDD churning). Then in games with the GPU loaded it uses about 200 watts. I only managed to get it to 233 watts when running both a cpu stress test (Orthos) and GPU benchmark (3dmark06) at the same time.

I'm in CA and pay about 13 cents/Kilowatt/hour so theoretically if I left my system on 24/7 (I don't) at idle it would cost me about \$11 per month. In reality I use my computer about 5 hours a day, with about half games (200 watts) and half just surfing, etc (120 watts) so say average 160 watts x 8 hours x \$.13 x 30 days = \$5 per month.
• #### glynor - Tuesday, November 18, 2008 - link

The prices listed on the EIA website that you used to generate all these numbers (for the US) are based only on the price residential customers pay for energy generation, and do not include the cents per KWH charge that all customers must pay for transmission. Even though I simply take the standard offer from my utility company (like most people), my actual price is roughly DOUBLE what the EIA lists as the average for my state.

Look at your bill. Make sure to add the totals for both Generation (Supply) and Transmission. These are billed separately.

With that in mind, most of the example totals reported in this article, and certainly all the comparisons to "Europeans paying more" are completely bogus!
• #### Maiyr - Tuesday, November 18, 2008 - link

Why bother with all of this just to figure out how much it costs to run your system ? Surely someone must have thought of just plugging in a Kill-A-Watt Electricity Usage Monitor....

Maiyr
• #### gochichi - Tuesday, November 18, 2008 - link

I would look for better efficiency rating for the following reasons (in this order):
1) I speculate the product is a higher quality product.
2) Inefficiencies turn into waste heat, cooler is more stable, more pleasant, and longer lasting (speculations)
3) I guess maybe saving a couple of bucks on my bill, but it's definitely not as important to me as just knowing that a have a "pimp" power supply.

What this article does make me want to do is try to figure out what sleep modes best suit my desktop computers (particularly my seldom used older desktop). Cold booting is inconvenient and I have been careless about finding a reliable sleep mode. I think the trick is to go S1, I just switched my main computer from S3 to S1 and it really helps stability and responsiveness.

I wonder how running laptops does in terms of power efficiency. There's got to be a ton of inefficiency caused by charging and discharging the batteries all the time. The design circling around a portable battery also makes it pretty power thrifty at the same time. Just curious as to how it actually pans out. Laptops have typically been more robust in sleep mode too though.

Random comment: Is it just me or have software updates gotten beyond the ridiculous point? Seems like they are not only often, but they are in your face... like it's OK to interrupt what you're up to a few times a week to get the update "right now"... I mean forcefully.

• #### bob11d50 - Sunday, November 16, 2008 - link

I just wanted to justify my purchases of energy efficient computer components.

The reason for this was about 3 years ago I had 9 people living in my house and the power company was on a teared system. I got up to the fourth tear where power was up to \$0.85 KWH. This was in San Jose California with PG&E as the power company.

I got my server down to 77 WATS with a EE Athlon at idle from about 230 with my dual Opteron. All measurements were using the Kill-A-Watt.

• #### sheh - Saturday, November 15, 2008 - link

I was always under the impression your total power readings were at the outlet, and so included the efficiency loss of the PSU. If that wasn't the case, how do you measure total power draw from within the computer?
• #### Christoph Katzer - Sunday, November 16, 2008 - link

Motherboard, system, graphics-reviews got readings from the wall. power supplies are loaded with a programmable load, so we can easily calculate the losses. You can know the power draw from mobos, gfx and so on when measuring the power distribution inside of the case which isn't difficult if you know which cable powers which part of which component ;) Reply
• #### bigsnyder - Saturday, November 15, 2008 - link

What kind of case is that in the picture? Reply