Why is it that the GeForce3 runs specific tests in 3DMark so well, most notably the Nature scene, when compared to other cards?

GeForce3 is the only hardware that implements all of the DX8 vertex and pixel shading instructions in hardware at full speed.

In the past, we have seen where disabling hardware T&L has increased 3DMark 2001 frame rates in the Nature scene. What's going on there?

That question is probably best explained by MadOnion. In order to truly test hardware T&L versus software T&L, you need to send the exact same data regardless of which box is checked. I think 3DMark 2001 sends less data when software T&L is checked. High-end CPU's are not faster at doing vertex shader operations than a GeForce3. We implement vector operations (SIMD processing) to a large degree, and in many cases, some of the vertex shader functions are many times the performance of the fastest CPUs out today.

Are we close to "needing" 128MB of RAM on graphics cards?

Absolutely! The more, the better! More textures, higher resolution textures, higher resolution displays, anti-aliasing—all of these are important, and cost more in terms of memory space.

There has been some public debate as to whether or not GeForce3-based products are DirectX 8.1 cards. What is the truth?

GeForce3 cards are DX8.1 cards, period. NVIDIA has Microsoft WHQL certified DX8.1 drivers for GeForce3. I believe that the new pixel shader versions introduced in DX8.1 do not offer significant new functionality over the original DX8 pixel shader versions; they are simply different. I'm not really sure what the value of these new versions really is. We certainly don't hear a lot of interest in them from game developers.

What, if anything, are we going to be missing by only having a GeForce3 DirectX 8.0 card compared to the competition's DirectX 8.1 card?

I am not aware of any picture that can be rendered on a competitor's product that cannot be rendered equally well, or better, on a GeForce3. Not only do I feel that the GeForce3 family products are the best performance products, they are also the most feature rich, stable and reliable products that I know of.

The only thing they may miss out on are compatibility headaches. As far as DX8 support goes, we licensed a ton of technology to Microsoft for DX8, so logic tells you it works. We are also the first to WHQL 64-bit XP drivers. We have been praised in the press for our stability in Windows XP, and Microsoft ran the computers at their XP launch events on NVIDIA GPUs. Think about it.

One thing that is drastically being underplayed in the press is the level of support for GF3 hardware in the developer community. Paper specs mean nothing. If they don't work, they are useless. If developers do not implement them and support them they are useless. Developers have been working on GeForce3 titles long before we shipped our first GF3 product in February. Some of our competitors are still trying to get their DX8 cards in the hands of the developers. The momentum we have there is undeniable. Why are there OpenGL extensions in games like Dronez that do not work on competitor's cards? Probably because we had cards to them when they were creating the applications. You can't code to and test hardware if you don't have it. I think more and more situations like this will rear their head as time goes on and more DX8 and OpenGL titles using shaders start to appear.

The TR staff would like to thank Dr. Kirk for his time. For more background and our take on many of these issues, please see our review comparing the Radeon 8500 vs. GeForce3 Ti 500.