Power consumption is essential to keep in mind when buying a video card. Despite its importance, this aspect of a video card quite often gets overlooked. The 8800 series are very high performers and so we expect them to be more demanding in their consumption.

• Motherboard: Intel D975XBX2KR (chipset Intel 975X)
• Memory: Corsair CM2X1024 6400 (3-4-3-9) @ 800 MHz; 2x1 Gbyte
• hard disk: Western Digital WD1600JS - Serial ATA - 7.200 rpm, 160 Gbytes
• processor: Intel Core 2 Extreme X6800
• PSU Tagan TurboJet TG1100-U95 (1,100 Watts)

We conducted two power consumption tests in two different states. In the first state, idle, we had an empty desktop with only Windows XP running while in the second state, full load, we had Sciencemark 2.1 running at a resolution of 2560x1600.

Looking at our tests, we can immediately conclude that the GeForce 8800 GTX demands the most power out of all other video cards. Under SLI, the 8800 GTS and GTX yet again consume the most power, while the Radeon X1950XTX ends up taking in nearly 100 less watts.

Test Configuration

In order to properly benchmark the new generation of video cards, it is necessary to use a test platform that will not create a bottle neck for the video card thus limiting performance. For this reason, we chose to test using the Intel Core 2 Extreme X6800, which is currently Intel’s fastest dual-core solution available for retail. We chose this processor over the QX6700 because current video games do not take advantage of four cores. The Core 2 Extreme X6800 also has a clock of 2.93 GHz compared to one of 2.67 GHz for the QX6700.

The motherboard we chose to test with was the Socket LGA 775 Intel D975XBX2KR based on the Intel 975X chipset. For memory, we used 2 Gigabytes of Corsair XPS CM2X1024-6400C3 running at a frequency of 800 MHz at timings of to 3-4-3-9. The hard disk was a 160 Gigabyte Western Digital Caviar WD1600JS based on the SATA interface and running at an RPM of 7,200.

For mutli-GPU SLI tests, we used the EVGA nForce 680i which, as the name implies, is based on the NVIDIA nForce 680i chipset.
We used Windows XP SP2, updated with the latest fixes available in October of 2006, for all of our tests. The driver version used for NVIDIA cards was Forceware 96.94 while for ATI cards it was Catalyst 6.10. For the tests with GeForce 8800 GTX and GTS cards under SLI we have used Forceware 97.02.

our test system wtih a Dell 3007WFP

A fundamental item that is required in order to analyze high-end video cards is a display that supports high resolutions. It would not make sense for a person to buy some of the most powerful video cards on the market only to use it for a low resolution display. Our choice fell upon two displays that natively support the resolution of 2560x1600 and are currently available for retail. We chose to use the Dell 3007WFP widescreen, which measures in at 30 inches.

In order to test these high-end video cards to their fullest, the resolutions of our benchmarks in this article do not get any lower than 1600x1200. We also made sure that under no circumstances anisotropic filter was set at anything lower than 16x.

For NVIDIA cards, we compared the GeForce 8800 GTX against the GeForce 7950 GX2 and GeForce 7900 GTX. The GeForce 7950 GX2 was chosen because it is one of the fastest single slot solutions from NVIDIA while the 7900 GTX was chosen because it is the main flagship for the previous generation of NVIDIA products. For ATI solutions, the main contender was the Radeon X1950XTX which was selected because it is the only contender to the 8800 GTX until R600 based video cards are released in early 2007. The 7950 GX2 board we used for testing was from XFX and was significantly overclocked to 570 MHz GPU and 1,550 MHz memory.

In order to provide additional information regarding the performance of these video cards when running on a 19inch widescreen LCD monitor with a maximum resolution of 1280x1024, we also ran some tests at this resolution.