It's time to move on from power testing most desktop PCs
Testing the power efficiency of most consumer desktops no longer makes sense. Here's why.
Perhaps you are aware that we conduct power consumption testing across a number of review categories.
I wish we'd started testing energy usage on desktops earlier, particularly with Intel's old power hog Pentium D chips. The problem is Moore's Law. Intel's Sandy Bridge chips (sorry, "second-generation Core") are too damn efficient. They're also pretty much everywhere.
The result is that our power draw tests aren't all that interesting anymore for mainstream PCs. The chart below helps illustrate the point.
All-in-ones and high-end gaming desktops
The dollar figure below to each system represents what we estimate you would spend to own and operate that PC for a year. (More on our desktop power testing methodology here.)
From the Dell XPS 8300 on up you'll find every higher-end gaming desktop we reviewed in 2011. Below the Dell, every all-in-one PC we reviewed last year.
The chart tells a pretty clear story. Power testing becomes academic when you find only an $8-a-year difference between the 20-inch, 1.6GHz HP Omni 120 and the 27-inch, 3.1GHz iMac.
Among the whole group of all-in-ones, 18 in all, the standard deviation for annual power costs comes out to a narrow $4.13.
Compare that spread with that of the 10 gaming desktops, whose $29 standard deviation suggests it might actually be worth measuring.
You can argue that anyone paying $5,000 for the Falcon Northwest Mach V can handle the $145 required to run it. Before you do, look at the Digital Storm ODE Level 3. You'll only pay $89 on the year to power that system. It also costs only $2,339, and offers similar gaming performance to the Mach V. On some game tests, the Digital Storm system is actually faster.
That kind of variability is worth tracking down.
We will check back in on mainstream power draw with big architecture changes, or whenever a vendor like Apple gets braggy about its eco-friendliness. But for standard all-in-ones, midtowers, and other nongaming desktops, we're removing power testing from our everyday benchmark gauntlet. That won't be the only change we make to our tests this year, so stay tuned, and feel free to share any suggestions below.
(Note: Our energy use calculations are based on the U.S. average residential monthly retail electricity price generated by the U.S. Census and published by the U.S. Energy Information Administration. We've continued to use our original 2009 figure of $0.1135 cent per kilowatt hour, but as of December 6, 2011, the average price is $0.1154.)