AMD Phenom II 965 X4 BE
2x2Gb Crucial Ballistix Tracer DDR3-1600
Swiftech H2O-220 Compact
Asus HD4890 Top
64Gb Patriot Torqx SSD
1Tb WD Caviar Black
Optiarc 24x DVD+/-RW
Cooler Master HAF 932 AMD Edition
I regret that I don't have anything more than a single HD4890 to really put a strain on the Sparkle 1250W, however as far as single-GPU cards go, the 4890 isn't exactly a power-sipping greenie weenie either. I'll be using the OCCT 3.1.0 Power Supply stress test to put a load on the Sparkle 1250W.
Idle and load measurements of watts used were taken using a Kill-A-Watt meter on the AC side and represent the wattage for the entire system. A power supply will typically reach its peak efficiency around 50% load and in fact that is where the Sparkle 1250W exceeds the 90% mark for Gold certification. However as you can see from this system we're only using around 10% at idle and roughly 33% at load. Even adding a second 4890 in Crossfire would likely still be below the half-way point.
As for how the Sparkle 1250W handles that load, well it does register a slight difference from idle. Using a Fluke multimeter and measuring at the back of the connector, the power supply shows a .01V to .02V drop going from idle to load on all rails. The +12V rail used to power the 4890 is the one monitored during the test. There was no fluctuation under load, meaning the readings stayed constant during the OCCT load test. Since this is such a small amount of drop and all of the rails are above specification to begin with this probably isn't an area of concern.
Let's wrap things up with some final thoughts and conclusion.