Installation was straight forward, and painless. The drivers supplied on the beta ASUS CD-ROM are version 3.48a, and have support for all the additional video in/out functionality. For this reason, I decided not to install the latest reference driver, but to stay with the supplied one. In previous testing, I have not found any performance improvement between 3.48 and 3.53 anyway. I also installed all the additional software for use of the Video input, and outputs, including ASUS Live, ASUS VideoSecurity, ASUS Tweak, and ASUS SmartCooling.

I used the ASUS Tweak utility to check the memory and graphics engine clock speeds, and was pleasantly surprised to find that the board was running at 193MHz memory/140MHz GPU. Obviously the Dynamic Overclocking technology allows the board to run at fairly decent speeds, as 193/140 is a fair bit quicker than nVIDIA's reference speed of 166/120. In previous GeForce benchmarking, I have noticed a decent performance improvement when the ram is overclocked from the standard 166MHz setting, so I would expect this board to be a strong performer straight out of the box.

After discovering this, I was keen to see how it actually performed compared to ASUS's V6600 'Pure' version. For further comparison, I have also benchmarked an AOpen PA3020 TNT2 Ultra card.

After parting with a serious amount of hard-earned cash, I figure that you'll want to crank the graphics options to the max, then find the highest resolution at which it is playable. After all, what's the point in playing Quake Arena at 640x480, 16 bit colour, all graphics options turned off, and getting 100fps? To take it further, I doubt that you are going to play with the sound turned off. I know that different soundcard drivers have different impacts on ultimate frame rates, but I would prefer to have some idea of what the overall frame rate will be, with a soundcard driver being used.

So, my benchmarking results were obtained using the following settings:

1. All graphics options turned up to the max
2. Sound card enabled

This way, you get a real-world type scenario, and a more realistic idea of what you should expect.

I tested the OpenGL performance using Id's excellent Quake 3 Arena 1.09 Demo001 timedemo, and tested the DirectX performance using Rage's Expendable timedemo.

Within Quake 3 Arena, I turned on all the graphics options such as ejecting brass, etc, and used trilinear filtering, 32 bit textures, lightmap lighting, and texture detail at the highest possible setting. I tested in both 16 and 32 bit colour. Here's what I found.

You can see that the higher default memory clock of the V6600 Deluxe gives it a reasonable improvement in performance throughout the range.

For comparison, I also tested at the standard graphics settings that are provided (High Quality, Normal, Fast, Fastest).

Don't try and compare these results with Quake 3 Arena 1.08, as the release of Quake 3 Arena 1.09 has meant a significant performance hit of around 10fps at the lower resolutions, and around 2-3fps at the higher end.

As with Quake, I enabled all graphics options, including bumpmapping. Here are the results.

You can see that the GeForce cards start to get a jump on the TNT2 at resolutions of 1024x768 and beyond, with the Deluxe version pulling away at the top end.