Performance Testing Methodology

System used:

Processor: AMD X2 3800+ (at 2.6Ghz)
Processor Cooling: Arctic Cooling Freezer 64
Motherboard: DFI Lanparty SLI DR Expert
RAM: 2GB Corsair XMS PC4000 DDR (at 520Mhz)
Hard Drive: Samsung Spinpoint 250GB, 7200Rpm SATAII
Case: Gigabyte 3D Aurora 570
Case Cooling: 3x Noctua NF-S12-1200 fans @1200rpm (1 intake, 2 exhaust)
PSU: Silverstone Decathlon 650W
Graphics Card: XFX 8800GTS 512MB Alpha Dog Edition

Coolers Used:

Stock Cooler @ default fan speed profile
Thermaltake DuOrb
Thermalright HR-03 GT w/ Zalman 92mm fan @ 1600RPM



We have installed the Thermalright HR-03 GT on our 8800GTS 512MB card and then installed the whole setup into a Gigabyte Aurora 570 case. All of the stock results are taken with the stock cooler in its default fan speed profile. All coolers (including the stock heatsink) were installed with Arctic Cooling MX-2 thermal compound. All of the tests were conducted with the side panel of the case closed.

The room where the test was conducted was kept at a steady 20.2°C (+/- 0.3°C) for the beginning of each test. Since we want to replicate real-world conditions, the temperature of the room was allowed to climb as the 30 minute tests were run. Before every subsequent test was run, the room was allowed to cool off to its original 20.2°C temperature.
For all of the performance tests we used nTune’s temperature logging feature to log the GPU temperature throughout each of the 30 minute tests.

All load tests were conducted by running a 30 minute loop of 3DMark 06’s Batch Size Test with 32,768 triangles. This test puts a constant and high load on the GPU and represents a worst-case scenario for any graphics card.

Any benchmark-specific methodology will be explained in its own section.