For some time now, the market has seen an all out explosion in wattage on the part of manufacturers. After 800 watts, several manufacturers made models available to the general public capable of supplying 1000, or even 1200 watts ! A new height in ridicule was attained at the CeBIT, when we saw first Enermax and then Thermaltake try and justify the use of such power supplies. This sparked our interest in power consumption and we decided to take this opportunity to find out where we are at in this domain. This will be the first part of a series of articles on the subject.1000 & 1200 watts ?!
Before moving on, let’s touch on the arguments of the above two manufacturers. With Enermax, there was a configuration with a consumption of 933 watts. Based on a Quad Opteron system, it had the specificity of having no less than a mere 24 hard drives.
Thermaltake went even farther, offering a configuration that consumed 1105.5 watts. The only problem was that only 605.5 watts were used by the PC, the other 500 was for an amplifier and speakers. These are peripherals that aren’t linked to the PC power supply.
In reality?
The idea of this test was, therefore, to know the actual consumption of our computers by changing various parameters with our ASUSTeK P5N32-E SLI Deluxe motherboard, 2 GB DDR2-800, Raptor 74 GB hard drive and X1950 Pro. Based on a nForce 680i SLI chipset, which allowed us to use the 8800 GTX in SLI, this card actually consumes a little more (20 à 25 watts) than its competitors based on Intel chipsets.
These first measurements were mainly done with the use of a wattmeter, which gives us figures directly from the wall plug. Our power supply, a 700 watt Seasonic M12, is 80Plus certified and, in practice, produces a supply higher than 80%, or in other words for each 400 watts consumed in 230V, 320 watts are provided on its 3.3V, 5V and 12V lines. We combined these measurements with those from an ampere meter, which gives us detailed information for the different lines.
We took a precise look at three configurations; a mid-range, high-end and then an ultra high-end, in order to see the specific power needs on each line. For this to be done, we loaded the graphic card and processor to their max. It’s still possible to have an additional 10-20 watts of consumption by permanently accessing the hard drive or DVD player, however, this type of test is difficult to put into place with multiple drives.
We also have to take into account that CPU & GPU load tests (Prime95 and 3DMark06) consume a lot, and it’s rare that programs use components in this way. Current use of many machines is closer to the load in standby. For example, on one of the computers while at rest we were at 129.1 watts, and we didn’t go beyond 150 watts while reading an audio file and surfing the Internet at the same time.
