Due to circumstances mentioned on the first page, I was left with a limited amount of time to test the GeForce 6600 GT. However, the benchmark results were extremely good for a graphics card that will be carrying a suggested retail price of $199.
I did get an opportunity to squeeze in about 45 minutes of Doom 3 gameplay. I started a new game using a resolution of 1024x768 with 2X application antialiasing using Doom 3's high quality mode, which enables 8X anisotropic filtering by default. The driver image setting was set to quality and trilinear optimizations enabled. With FRAPS logging enabled, I ended up with an average of 58 frames per second and a minimum 27 frames per second. Very impressive results indeed.
GeForce 6600 GT Logo
The image quality associated with the various texture filtering optimizations in the 65.76 drivers is quite good. With custom profiles and the ability to control the degree of optimizations allows one to fine tune image quality and performance down to the game level.
One potential drawback with the GeForce 6600 GT is its 128-bit memory bus, which will limit antialiasing performance at high resolutions. An example of this bottleneck can be seen in the Enemy Territory benchmark results.
Benchmarks consist of Unreal Tournament 2004, Halo, Enemry Territory, and Doom 3.
UNREAL TOURNAMENT 2004
The Unreal Tournament 2004 benchmark results shown below were not intended to be a direct comparison of performance between two different systems. The results were intented to illustrate how each of the graphics cards responds to texture filtering optimizations. A chart showing the relative differences in performance could have been used.
The built-in Halo benchmark was used with default graphics settings, which include high texture quality, specular lighting, shadows and decals. Particles were off.
Halo - No AF and 8X AF
FRAPS was used to determine the minimum and average frame rate while playing back a demo at normal game speed from a 6 vs. 6 player match that takes place on the Railgun map. All in-game graphics options were set to their maximum value, texture compression was enabled, and sound was disabled. Performance was based on the first three minutes of the demo.
As mentioned in our Doom 3 Guide, benchmark results are based on an average of the second and third benchmark runs. Results from the first run were not used. The minimum frame rate was obtained with FRAPS.
This following is a list of changes that have been made to this preview since it was originally published:
September 7, 2004 - Page 1: Added a link to a DDR vs. DDR2 memory comparison at Overclocker Cafe. Mentioned that the GeForce 6800 Ultra was used in putting together the image quality comparisons in this article.
September 7, 2004 - Page 7: Added the GeForce 6600 GT logo, added information about each benchmark, corrected graph labels for Doom 3, Halo, and Unreal Tournament 2004.