Recently when we've tested power supplies, we regularly received comments about the fact that we test the efficiency of PSUs beginning at 200 watts. That's because the energy consumption of modern desktop PCs with Ivy Bridge or Haswell processors is typically a lot lower. It's a valid point, so we spent a lot of time in retesting many of the PSUs we've reviewed in recent months. This time we tested their efficiency at much lower load levels, of 22.5 watts, 50 watts and 100 watts. The results are definitely worth taking a closer look at.
Our procedure for testing PSUs has been the same for a long time now. We test power supplies in increments of 100 watts, starting at 200 watts. We test a 600 watt PSU at 200, 300, 400, 500 and 600 watts. At each step we take 50 watts from the 3.3 and 5 volt lines, with the rest coming from the 12 volt rails.
That's different than how most other hardware sites do it. Most follow the 80 Plus method of testing at 20 per cent, 50 per cent and 100 per cent of the maximum capacity, dividing the load the way the manufacturer has envisioned. Our approach makes it easier to get comparable results, and by limiting the load on the 3.3V and 5V rails it corresponds more closely to the real-life situation of PCs. The two most power-hungry components in a computer, the CPU and the GPU(s), only take their power form the 12V rails.
We've now added three more tests with loads of 22.5 watts, 50 watts and 100 watts. These levels and how we distribute the load over the 3.3, 5 and 12 volt rails is based on an actual PC. That system consists of an Intel Core i7 4770K Haswell processor, a Gigabyte GA-Z87X-UD3H motherboard, two Corsair Vengeance Pro memory modules, an OCZ Vertex 4 SSD and a CPU cooler. We used an ammeter to get a reading of the system at different points, and measured the level on each cable coming from the PSU.