The Max settings are attained by using Quake 3's default high quality settings, and tuning it to get more eye candy from a higher resolution, high geometry, trilinear filtering and high quality textures. The tests were run at 1024x768 with 32 bit color, which happens to be the setting at which I play :)

The real world results speak for themselves. In a scenario in which the CPU is working as hard as it can to set up the pretty polygons being displayed by the GPU on the monitor, faster RAM settings definitely improve the game's performance. When the game is limited by the performance of the video card, as is the case here, a faster RAM settings do not have as much impact as one would hope for.

3DMark2000, is the essential Direct3D benchmarking program. MadOnion has cranked out a solid metric built specifically to test the Direct 3D performance of your system.

Here's how the Mushkin stick fared:

And for the sake of setting the results up in a similar fashion as the previous charts...

I ran the default benchmark, which tests at 1024x768@16bits at both fsb speeds once again. The Mushkin memory running the Cas 2 setting comes out on top once again. There's no fooling the results of this benchmark as the improvement is almost 300 3D Marks at the 133 MHz fsb speed by going to Cas2 versus Cas3. At 100 MHz, it's just over 200 3D Marks. This is the first and only result which suggest that running Cas2 at a higher fsb speed results in a better performance increase than at a lower fsb speed.

Although I'm only examining the final 3D Mark numbers the program churns out, you can examine the results of every test by examining the following thumbnail.

How high can this RAM really go?