Efficient power supplies are actually efficient!

I have heard a lot about the new more efficient power supplies, and I was eager to compare for myself but at the same time too cheap to buy a new one :) Luckily (for me only) a friend’s PC died all of a sudden, so he kindly gave me his fancy NZXT 650 watts to replace my generic. old 550 watts PSU. The new one was a “80 plus bronze” rated PSU.

The new PSU went into my server at home. A dated machine with AMD Athlon II X4 635, internal graphics, four 3.5″ Hard drives and an optical drive. This machine does quite a few things for me: CUPS Print server, DLNA Streaming, Samba sharing, Development web server (Apache for PHP, Tomcat for Java, MySQL and PostgreSQL), and I am probably forgetting other things. Although this is not a scientific experiment, I tried to keep things a bit scientific by changing nothing in the machine’s configuration but the PSU.

I had attached this machine to my power meter for a few weeks before and after swapping the PSU. The difference was quite clear as outlined below:

  • Running idle (not sleeping): 90 watts old vs. 72 watts news. 20% more efficient.
  • Under heavy load: 115 watts old vs. 85 watts new. 27% more efficient. Note: Heavy load involved more disk IO than CPU usage – it is pretty hard for me to max out those four cores!
  • Maximum load recorded: 150 watts old vs. 134 watts new. 11% more efficient. These number were typically scored during startup.

Considering this systems runs idle or almost idle most of the time, the total consumption saw a 20% overall drop (recorded over a few weeks).

Although that sounds like a lot of electricity, unfortunately, that does not translate as well into a dollar amount because electricity is still quite cheap (at this scale of consumption). At the current rates where I live, this translates to savings of $10-$12 per YEAR. In other words, if I had to buy this new PSU instead of getting it for free, it would have taken six years for the “investment” to pay off.