There are smaller textures in Game 3 test: only 64MB is required. However, the vertex and index buffers increase to 19MB and 2MB respectively. Anyway, this doesn’t affect the final results: GeForce 6600 GT and ASUS Extreme N6600GT Silencer run equally fast here. Just as in Game 2 benchmark, overclocking allows ASUS’ solution to win the leadership in all resolutions and in all test modes. As we can see, Game 2 and Game 3 tests feel quite comfortable with 128MB of graphics memory available even with FSAA and anisotropic filtering enabled in high screen resolutions.
Game 4 test sets more sever requirements for the graphics accelerator: the overall amount of graphics memory allocated for textures, vertex and index buffers reaches 113MB, and theoretically, the test may require over 128MB of memory in high resolutions. This theory comes partially true, because ASUS Extreme N6600GT Silencer starts to outperform GeForce 6600 GT in eye candy mode in all resolutions above 1024x768. the performance gain we observe is about 10% in 1280x1024 and grows up to 20% in 1600x1200. In all other cases the card performs just like any other GeForce 6600 GT. Overclocking makes our hero an absolute leader only in eye candy mode, while in pure speed mode the overclocked ASUS Extreme N6600GT Silencer runs almost as fast as RADEON X1600 XT.
Well, as we have just seen, our suppositions about the performance improvements of the ASUS Extreme N6600GT Silencer have just been proven true: only in two tests 256MB of graphics memory appeared really demanded, and that was only in heavy test modes. In all other cases the ASUS solution performed just as any other GeForce 6600 GT would do.