Setup & Results
To test power supplies, we believe that creating real-world circumstances and conditions are important to consumers who need to relate to what they would use and experience in a computer system setup. Consequently, our methodology is fairly straightforward: we assemble the most power-hungry components we can muster, hook up the power supply, and take measurements at idle and load using a Kill-A-Watt meter for wattage, and a digital multimeter for voltage output and fluctuations.
To break this down a bit further, we conduct primarily 4 tests:
- Idle: The system sits at idle Windows
- CPU Load: In this test we run concurrent instances of Prime95 on all CPU cores
- GPU Load: In this test we run FurMark, as it recognizes SLI and CrossFire setups and loads the GPUs very heavily with maximum settings
- System Load: In this test we run concurrent instances of Prime95 and FurMark to draw the maximum amount of wattage possible
The BFG EX-1000W is a monster and with that much available power it may be difficult for most users to muster something that can push it that hard. However, we’re pulling out all the stops today, as we’ll start with an Intel Core i7 quad core processor that is highly overclocked. It is cooled with a exceedingly large thermoelectric (TEC/Peltier) watercooling setup that also draws a massive amount of power on the system, far more than any traditional cooling method. We then connect two dual-GPU cards in Crossfire/SLI to tax the power supply even further. This setup can easily pull down some extremely high wattage, pushing the EX-1000 to its limit, and far more than even the most hardcore systems that users would have at the best of times in a real-world setup.
When examining the results we are looking for the voltage output and any fluctuations that might occur. The current ATX specifications allow for the following fluctuations in voltage outputs, and these represent a 3% variance:
- 3.3V Rail: 3.135V – 3.465V allowable
- 5V Rail: 4.75V – 5.25V allowable
- 12V Rail: 11.4V – 12.6V allowable
Below are the results:
When we load up a power supply to 100% then we start to see the winners separate themselves from the contenders. And this is where the EX-1000 starts to struggle just a wee bit; specifically in the +12V regulation. While the results are indeed with acceptable standards, the 0.20 drop is a bit more than we’d like to see on a kilowatt unit that would likely be powering some very expensive hardware. There is no cause for real complaint though, as this is within spec, it’s just more of our preference from what we’ve seen through many tests of products.
Regarding the noise produced by the unit, as we know, the higher RPMs a fan speeds, the louder the noise, but then also the cooler the temperatures. It is an ongoing balancing act between noise and cooling for just about any mechanical component in a system. To that end, it seems that the EX-1000’s fan tends to run at low RPMs when idle and the fan is very quiet. When heavily loaded, however, the fan doesn’t ramp up much and is still very quiet. While this is great to alleviate any noise concerns, it comes at another cost: heat.
The heat produced by the EX-1000 when fully loaded is rather significant, which is likely a result of the small heatsinke we noted when we looked at the interior of the unit. This power supply does get very warm when pushed to its limit for long periods of time. Granted, most people likely won’t stress a kilowatt unit such as this as hard as we do during testing, but some might. Unfortunately due to the cramped internal layout, there isn’t much more that can be done to significantly increase the surface area of the heatsinks to try to reduce the heat being produced.