NV30 = GeForce FX
The Dawn of Cinematic Computing

Article written by Hilbert Hagedoorn
Date: November, 2002

A few days ago some members of the press where asked if we would like to receive a little insight on NVIDIA's upcoming graphics processor. The codename is NV30 and this silicon is of course the immediate follow up of the NV25/28 (GeForce4 Ti). This article is a preview based upon info that I received from NVIDIA last week. Some pieces will be hard to understand some of them shall be easy to understand. Let me say it immediately, at this moment NVIDIA has not distributed samples to the press, so do not expect benchmarks just yet. As always we where given this info a tad late, so a lot of the excerpts of this article are in fact directly taken from NVIDIA's technical documentation. We however will give you an insight of what to expect of the new silicon that has been named ... GeForce FX.

Yes, ladies and gentlemen, GeForce FX is finally upon us and we are one step closer to rendering games and 3D scenes as you know them from animated movies like Toystory .. we heard and have seen many rumors regarding this new NVIDIA silicon. Rumor is that it took NVIDIA 400 Million Dollar to research and develop it and fact is that NVIDIA took some very risky steps to insure themselves this product will be successful, it is the single to largest investment they've ever made. That by itself makes a strong statement, this is a completely new generation silicon with a very unique concept and a big step into the future of 3D gaming. It's where NVIDIA combines the 3dfx legacy and make the mojo that is called GeForce FX.. Funny to see are all the names people thought this product would have, I heard names like: GeForce5, Nitro, Eclipse to the name that is now official: GeForce FX

This is her, the NVIDIA GeForce FX reference design

The Product

After reading the confidential material I figured I should immediately start of with some words that should shed my overview of ideas and thoughts on NV30, let me try to get you an idea of this new GPU in one line:

The GeForce FX Graphics core will be the fastest, highly programable and most feature rich GPU that can render games with cinematic detail & precision.

The first thing you guys always want to know though is performance. Performance wise it should be about twice as fast compared to a GeFore4 Ti 4600, it has a very efficient memory architecture combined with high-speed memory, way better Pixel and Vertex Shader support and it is a programmers dream as this is a very flexible card to program. GeForce FX will be a highly configurable and programmable chip with the help of Cg. The Cg language has been launched and designed with GeForce FX in mind. The developers willing to use the features of GeForce FX to the full extent will have to resort to Cg and can unleash little wonders.

Saying it simple, the GeForce FX is going to kick some butt, that's a fact. NVIDIA is building some strong marketing as we speak in calling it the new GPU 'The Cinematic Shading GPU' as this graphics card can render games with film-quality precision. First we are going to walk trough some of the Specifics and then a little in depth. This article will have several chapters.

This new GPU is and will be profiled as Cinematic GPU as it is capable of bringing cinematic visual effects on your PC with the combination of some brutal power and an excellent featureset. The CineFX GPU is of course a capable of utilizing DirectX9 Pixel Shaders 2.0+, Vertex Shader 2.0+ and OpenGL.

Such a GPU of course deserves good memory bandwidth. That rumor was true .. the GeForce FX GPU will make use of the very excellent DDR2 local memory which is capable of a 500 MHz thus effective 1 GHz frequency !! And although nVIDIA did not announce the core clock frequency yet you can expect it to be in the 400-500 MHz ! In theory you can do 20.000-22.000+ 3D Marks with a graphics card like that ..

We can always get a vague idea where a card will be performance wise by looking at it's memory bandwidth. Based on the DDR2 frequency we can make note of the fact that  [(2x128bit) x (1000MHz:2) : 8bit] the graphics card has roughly 16000 MB/sec memory bandwidth. It's not all about memory bandwidth of course.

Realtime Cinematic Effects

Developers on GeForce FX - John Carmack (Id Software)
NVIDIA Is the first of the consumer graphics chip companies to firmly understand what is going to be happening with the convergence of consumer realtime and professional offline rendering. The architectural decision in the NV30 to allow full floating point precision all the way to the framebuffer and texture fetch, instead of just in internal paths, is a good example of far sighted planning. It has been obvious to me for some time how things are going to come together, but Nvidia has made moves on both the technical and company strategic fronts that are going to accelerate the timetable over my original estimations.

My current work on Doom is designed around what was made possible on the original GeForce, and reaches an optimal implementation on the NV30. My next generation of work will be designed around what is made possible on the NV30.