FSAA and Shane Long

FiringSquad: We originally reported on the inclusion of a 2-sample anti-alias feature that would be faster (but uglier) than the standard 4-sample T-Buffer effect. Recently, we've heard that 2-sample AA would not be included. Can you update us on the situation?

When we launched at Comdex we announced 4-sample AA for the Voodoo5 and no AA for the Voodoo4. I don't have anything new to tell you today…

FiringSquad: We haven't heard much from 3dfx lately regarding FXT-1 texture compression. Are game manufacturers beginning to support this new open-source standard?

It is receiving a lot of support from developers, especially in the open source community. Our developer relations team has done a great job of evangelizing its benefits. Of course, they will have a lot more leverage when VSA-100 ships and there is hardware available to support it.

FiringSquad: How has Shane Long's departure affected 3dfx business development? Does it lend credibility to Bitboy's Glaze3D (or its own impending launch?)

Shane made many contributions to STB over the years, we'll miss him and we wish him all the best. We've reorganized our sales organization under veteran retail market leadership that's more appropriate for our business model as we move forward.. I think that Bitboys made a good move in adding Shane and his OEM experience. We always take any competition seriously, but credibility in the 3D graphics business comes from only one thing: delivering real products.

FiringSquad: The Glaze3D claims to achieve much of its speed through the use of embedded DRAM. On a similar note, both Intel and AMD are moving sections of their memory infrastructure onto the CPU die. What is your impression on EDRAM? Is on-die cache/memory a trend or an interesting technology offshoot?

I don't think 3dfx is ready to officially line up on one side or the other. We are looking very closely at embedded memory technologies and ways to get around their current drawbacks. It really increases die size and with that comes increases in cost. Anytime AMD and Intel put their stamp of approval on a technology, I think it should be taken seriously. So far we've seen no reasonable solution to the problem of compromising logic speed and density in order to achieve reasonable speed and density in the EDRAM.