by Leonard "Viking1" Hjalmarson

  My guess is that by the end of next year this will definitely be the case, but without hard performance figures on this years it is difficult to say. We are not overly concerned about the flexibility of the hardware, firstly because the first spin seems to match well to what we are trying to do, and secondly there is a strong drive to free this up over the next couple of years. Certainly the plans for Direct X 8 and beyond include a large measure of programmability for the hardware.

Higher fill-rate is desirable, but not the holy-grail. I think T & L hardware is the way forward. 3Dfx's accumulation buffer scheme is an implementation of a good concept from the Open GL world, but is more a marketing gimmick than anything else. We can do effects with it which we can't with other cards, but the same is true of PowerVR's shadow volumes.

Julian P H Davis (Jules)
Principal Software Engineer
Kuju Entertainment Limited. (Team Alligator)

Tucker Hatfield

Basically, I tend to look at new hardware features and get excited about what we could do with them. Hardware T&L will be very sexy. Think of all of the cool things that could be done to add a higher level of realism to sims with this. Look at Rowan's MiG Alley, and notice the things they are doing with light and shadow, with highlights and reflections, even the glint you get from an aircraft at a distance.

This has some pretty high coolness potential but it still looks pretty rough. This is no slam on Rowan... if they had had access to beefy T&L engine, they could have made those effects sing and dance and still have had CPU cycles to burn. And think about complex lighting models combined with bump mapping and you can conjure up some pretty amazing images in your head...

Click to continue

FS 2000

However, having said that, while I think that hardware T&L is sexier, when I put on my gamer hat I think most of us will get the most, quickest benefit from improved fill rates. Why?

Well, to make the most of T&L hardware developers have to come up with a lot of cool ways to implement the feature set. It will mean that it will be a while after the cards come out before we see radical improvements due to the T&L hardware. A card that implements increased fill rates will almost certainly give noticeable improvements even to existing games by taking some burden off of the CPU, improving frame rates overall.

For games designed from scratch to take the most advantage of the fill rates, that gives lots of extra CPU time to devote to other improvements in software. This is likely to give us the most immediate and noticeable overall improvement in sim quality, since the extra clock cycles can go to AI, additional graphics, etc. If fill rates were radically fast, we might even find that T&L hardware wasn't necessary...

In short, I think T&L hardware will deliver some potentially amazing, new features in sims, but faster fill rates and full-screen anti-aliasing will give us the fastest and broadest improvement in overall quality and frame rate.

Tucker Hatfield, Microsoft FS 2000 Team

Go to Part III