Test Setup and Methods
Test Setup

• Case: Vigor Force (3x 120mm Antec LED fans @ 1200RPM)
• Power Supply: Cooler Master Real Power Pro 850
• Motherboard: Intel Desktop Board D975XBX2 (Bad Axe 2) | ASUS P5N32-E SLI (680i)
• Processor: Intel Core 2 Extreme QX6800
• Hard Drive: Western Digital 2500KS 7200RPM, 250GB with 16MB Cache Buffer
• Video: NVIDIA GeForce 8800GTS 640MB (Reference board) | (2x) NVIDIA GeForce 8800 Ultra (Reference board) | (2x) XFX NVIDIA GeForce 8800GTX eXtreme (600/1900) | (2x) ATI Radeon HD 2900XT 512MB | (2x) ATI Radeon HD 2900XTX 1GB | (2x) Diamond Viper Radeon HD 2900XT 1GB
• Memory: 2048MB (2x1024MB) Corsair XMS2 PC2 8500 (running at PC2 6400)
• Optical Drive: Lite-ON SHW160P6S05
• Cooling: Vigor Monsoon II
• Audio: Creative X-Fi XtremeMusic

Software Configuration

• Motherboard BIOS: D975XBX2 Express BIOS Update (Rev. 2214) | P5N32-E SLI BIOS Rev 1002
• Operating System: Windows XP Professional with Service Pack 2
• Video Driver: NVIDIA ForceWare Version 158.18 (May 2 Relase) | ATI Catalyst 7.5 (Diamond Viper cards), ATI Catalyst HD 2900XT 8.36 RC4 (others)

We were lucky enough to get the opportunity to test some of the 512MB cards a long, long time ago, and from what we saw (and felt) from that initial experience was that the heat generated from these cards was so immense that it was clear they were consuming (and consequently leaking) a lot of power. Bearing in mind this was before AMD/ATI publically (or privately for all we know) released any kind of power consumption figures. We later found out that they consume 215 Watts, which seems reasonable based on testing we have done with a Kill-A-Watt meter plugged into a CrossFire system. As such, we chose to use an 850W power supply in the system, as it would still be adequate.

You might also notice the hodgepodge of cards and drivers that were used in this test. This is the result of the full extent of this test being spread out over a period of about two months, with actual testing occurring for about 5 out of those 60 days. Driver versions that were available and provided to us were far from uniform, with release candidates and the like in abundance. Futhermore, you will notice when we get to the meat of this review that there are many tests that do not include the same sampling of video cards. This is because several of the benchmarks were not available to us while testing certain cards, and certain cards were not available to us when using certain benchmarks. It is an unfortunate consequence of jumping on these opportunities to review cards before we are really prepared to, and we sincerely apologize for this. However, to make it up to you we will be publishing a full review of these cards in our entire benchmark suite sometime in the future. For now, we feel that what we have assembled here is more than comprehensive enough to base decisions off of.

Our testing methodology of averaging at least three tests for each number in the results applies to this review. However, due to the limited time and the extensive periods in which no testing was done, some aspects of the tests were not carried out exactly the same every time. All quality settings were always consistent, but you may notice that some of the resolutions are different. Please be weary of these discrepancies, as they may well be causing some really weird looking numbers. Also, if at any point we discovered some sort of anomalous data point in testing, we will make sure to point that out. All of these products were brand new and essentially untested when we got our hands on them, so random errors and incompatibilities were abound.