First I used the ol' multimeter to see what kind of voltages I would get on the +5v and +12v rails, which were 5.01 and 12.01 respectively. You don't get much better than that.
Next, I launched OCCT 2.0. OCCT is a program that stresses your CPU and memory. The cool thing about it is that while it is benchmarking, it charts various temps and voltages in your rig, including +12v. At idle, the +12v shows 12.016.
The first test I ran is a 30 minute test. During the first minute and final four minutes of the test, the system is at idle. During the remaining 25 minutes, the CPU runs between 98 - 100% load. I left the multimeter connected during the testing, to make sure the software reflected the actual known voltages.
Once the stressing actually started, there was a lot of flutter in the +12v, but if you will notice, the fluctuation was only 6/100ths of a volt. The multimeter never changed from the 12.01, and it wouldn’t since it doesn’t measure in hundredths. Since the industry standard is +/- 10%, which means a fluctuation from 10.8 to 13.2 volts is allowed, the flutter we saw during the testing is negligible.
Next, I ran a 10 minute test, with the first minute and the final four minutes running the CPU at idle. At three minutes into the testing, and again at six minutes, I launched Ozone3D’s FurMark, an OpenGL benchmark that puts the GPU under a 100% load. Since the video card draws more power than any other part of the system, I thought that it would be interesting to see how a full GPU load would affect the EliteXStream.
During this test, during the time the GPUs were at load, the +12v dropped the same .06v, and as soon as they came back to idle, the OCZ800EXS performed as before.
During all testing, the EliteXStream was silent, and was blowing barely warm air.