Our load tests utilize a couple of FAST ATE active load testers and a variety of other equipment such as an oscilloscope, power conditioner, temperature probe and a power consumption meter. You can read more about our standard testing approach here.
The tests performed are based around six conceivable setups that are out there and progressively load down the PSU up to the power supply's limits or 1000W, whichever comes first. Since our test equipment's limits are equal to the output of the Rosewill RGD1000, we can test it to the maximum.
Initial testing proved promising, but things went downhill rather quick with the RGD1000. The 3.3V and 5V rails just don't manage the load quite well and get close to being out of specifications by Test 4.
In Test 5, the 3.3V rail drops all the way down to 3.05V and the 5V rail is at the limit. Pushing the RGD1000 to the max shows the 5V rail also dropping below ATX specifications. Efficiency, however, is right where it should be to support the 80Plus Gold rating that the unit has received.
As there is no specific test at 50%, it is impossible for us to tell if it hits the 90% efficiency as required, but even if it doesn't, it is very close.
Rosewill offers quite a bit on paper with the RGD1000. Superficially it appears to be everything you could want in a high wattage power supply. It promises plenty of connections, a 50C rating, 80Plus Gold efficiency, and all the voltage and current protections you can get. Even upon opening the box there were high hopes for this unit, as it appears of quality and would look good in just about any rig.
Unfortunately the RGD1000 fails to deliver where it matters most. While it isn't uncommon to see high wattage power supplies border on failing tests at the maximum, the RGD1000 showed signs of sub-par voltages at around 70% load. As you will see in some of the upcoming PSU reviews here at TweakTown, there are simply other power supplies that perform better than the RGD1000 for your money and that are in the same output range as this Rosewill unit.