There is indeed an end to the usefulness of 3D accelerators, although it's not here yet (and I don't think it will arrive anytime soon). But I do agree about the resolution case. After my PowerVR 250 article (in which VideoLogic claimed that SLI, PGP is not needed), I recieved a lot of emails telling me that Videologic should support PGP because the gaming community wants to see games running at 1600x1200. I want to make one thing clear: SLI is a 3dfx-only technique, PGP or Stepsister technology is what Metabyte is doing. With SLI the two boards take turns rendering one scanline after another on the monitor. Metabyte has another approach with PGP. PGP works by dividing the screen into 2 horizontal parts. One card renders the top half of the screen and the other card renders the bottom half of the screen.

Videologic's claims about PGP could be justified, but I think there is a more complex issue at the heart. How many of you can run games at 1600x1200 at 85Hz or better? How many of you use a 19 inch or higher monitor? The latest polls on AGN3D show that less than 5% of people have a 19" or better monitor. I think 1024x768 at 32-bit, at a minimum of 60FPS (I prefer 90FPS), is something we should go for this year. SLI showed what it is to run a game at 1024x768, but image quality was lacking with the Voodoo. What did not happen was getting the lowest scores in games on a higher level. Unreal still drops to 30 - 35 FPS with SLI (on a PII 450). I would like to see games that run at a minimum of 60 FPS, even on the most complex scenes. Things such as large textures will give a performance hit, and whether 3dfx likes it or not, larger textures than the Voodoo 2 or 3 can handle are going to be used in this year's games. It may look like we are anti-3dfx, but this is not the case. The Voodoo 1 and 2 got us where we are in 3D gaming today, but when the company who was on the leading edge of 3D acceleration still insists on using old techniques, then there is no evolution. We need progress!

Speed is not everything. Image quality and features are equally important and 3dfx's claims are very nice, but NVIDIA (with the TNT 2) now proves that you can get good speed and good image quality together. Still, if you want the highest frame-rates possible, you can't go wrong with 3dfx. The Savage 4 and PowerVR 250 also include texture compression and some other very special features. AT CeBIT '99, Matrox demonstrated their hardware bumpmapping demo to me, and I was quite impressed.. Bump mapping will be included in many future games and new videocards should support it. This year we should all be playing games at 1024x768 at 60FPS or more with a lot of great new features. It does not matter if you buy the PowerVR 250, Voodoo3 or TNT 2. They are so close to each other in terms of performance, that any card will be a great asset to your system. S3 will be somewhat slower, but performance is expected to be good as well.

So where does this all lead? Well I predict that within a couple of years there will be an end to polygons. There may also be other solutions to this dilemma. Curved patches, which are planned for Quake III Arena, (although Carmack may add support for non-rational bezier patch nets so that NURBS can be imported later on) is a step in the right direction, but it is not the solution. The first game to feature NURBS is Prey. Voxels maybe? We will see. Carmack has already said that voxels are a good solution, but we need some kickass hardware if we want to use them. A gig of ram, next gen DVD, you name it. 5 years is what Carmack thinks we need before we use voxels. 3D cards will also need to support voxel acceleration.

Before 3D accelerators became really important, there was that hype about virtual reality. I still think that virtual reality is the future. When 3D accelerators hit the end of their performance level, the next logical step is VR. Imagine going into a VR deathmatch on the Internet with Quake V! And what about when you play games online and you can actually talk to each other? There are great things ahead.

Well, not everyone may agree with some of the things stated in this article, but that's good. We need discussion about 3D chips and where we want to go with 3D gaming. This keeps creativity alive, and that is what we need. There are some games out there that are pure demonstrations of 3D technology (Forsaken, Terracide, etc). When you do not agree with our articles at Beyond 3D, please feel free to make comments or ask questions (as many do).

For now I really can't wait to test Quake III Arena on the PowerVR 250, Savage4, TNT2, Voodoo3 and any other card I have around here.