|
|
|
How long did it go on like that and what's the deal with Mr. Smith and squids?
About 6 months - between the time when we incorporated and opened up our first office. You'll have to ask Ross about the squids.
What sort of items fell onto the Voodoo Graphics plate through your work with SGI/Unix based systems?
Pretty much everything in the OpenGL pipeline. I designed it to be a superset of the OpenGL pipeline, but also left out a few things that didn't make sense for gaming at the time. I have the OpenGL state machine poster right next to my monitor on my desk.
How did that differ from the approach taken by previous industry leaders?
Well, there were no industry leaders in 3D in 1994. Many competitors skimped on quality, the established 2D leaders more than new 3D startups. Ditto for performance. Ditto for features. I think history has shown they paid a heavy price. Many competitors asked themselves how much 3D can I do within a given budget (dollars or gates - all the same thing). We asked how many gates or dollars will it cost to do what we want. They underestimated how much people would pay for good 3D, we aimed high and waited for costs to come down.
What we're wondering here is how in heck you managed to put out a chipset in 1996 that was so far beyond the competition that nobody even came close for many moons?
Well we left out 2D! This simplified the problem and freed up lots of gates. And we used 2 chips. Totally wacky ideas that many thought were crazy. That coupled with hard work, focus, and a determination to deliver the best 3D possible.
Did you purposely leap-frog your competition's technology as part of some ingenious business strategy or was the whole thing just a fluke?
It goes like this: we aim high, higher than where our competitors are saying they are aiming. We deliver, they don't. Therefore we not only beat their hype, but we beat their delivered product. So beating hype is part of our goals. I wouldn't say it's our only business strategy, but having a technology lead is about one-third of our strategy. Beating their delivered products is not a fluke, it's being fortunate that your competitors stumbled. We don't expect them to always stumble, but are happy and thankful when they do.
We seldom talk about where we are aiming - and there's a good reason for this. It's very difficult for our competitors to take aim at us!
With so much open ground to play in (Rendition wasn't exactly creating a new paradigm for 3D acceleration) how did you go about choosing which advanced 3D features for your first product (Voodoo Graphics) to implement and which to leave for a rainy day?
We talked to developers, conducted a survey at the first CGDC we attended, and used common sense and the knowledge we had accumulated working in 3D in the past.
You obviously had to make some technology trade-offs. Are there paths you didn't choose that you would do differently with hindsight?
We almost didn't include z-buffering because memory prices were so high, and developers weren't sure they would use it. But, we did it just to be safe, and sure were glad we did! The very first chips did not include an 8-bit palette for textures, a format very commonly used by developers. These chips never made it to market. We corrected for this in our second chips, which were the first to be shipped in a product. Some of our competitors still don't natively support this texture format.
|
 |
 
|