It was painful to watch 3dfx slip from the archetypical kick-ass technology start up to where they wound up.Not terribly ceremonius, but then, he was on the company's advisory board and got to participate in frustration.
I think I would have been happiest to have the PC market divided up between three strong players that all had their act together, but at this point, I'm not too unhappy with the market simplification resulting from 3dfx exiting.
Then there's Unreal lead programmer Tim Sweeney's comments over at Evil Avatar. He seems a little more melancholy about the graphics pioneer's passing. But he did pee on the grave of the one 3dfx technology that seemed (to me) most likely to survive the NVIDIA buyout: Gigapixel's tile-based rendering technology. Get this:
NVidia needs to bury the whole Gigapixel chunking architecture they're inhereting. It was a horrible idea. Capturing all your geometry and buffering it up in RAM for later rendering in little chunks was a horrible idea when VideoLogic ran it into the ground; it was a horrible idea when Gigapixel ran it into the ground; and it was still a horrible idea when 3dfx ran it into the ground. Let's hope NVidia kills the damn thing once and for all.I hadn't seen such a strong statement against the seemingly more efficient tile-based approach in these days of memory bandwidth crunches. (Thanks to the 'shack for those two links.)
Anyhow, that's not really the point, I suppose. The thing is, 3dfx is yet another company with pioneering technology that created a market, then couldn't sustain success in that market. It's sad to see, and all the more so because the story is so familiar. Since the Voodoo 3, watching 3dfx has been like watching a car accident in ultra-slow motion. (And in 16-bit color. Heh.)
A number of you have speculated in the comments here about how NVIDIA will make use of 3dfx's technology to better compete with ATInot an uninteresting question, in light of the rumors we've heard about the 3dfx Rampage/Spectre. But honestly, I don't think NVIDIA is going to make use of 3dfx's technology at all. They might find the Gigapixel stuff useful for PDAs and cell phones, but that's about it. NVIDIA bought 3dfx to get rid of a lawsuit, to own some patents, and to eliminate some competition. As for the 3dfx technology, I don't think there's much in it for NVIDIA. The GeForce core is already more advanced than anything 3dfx has produced, and NVIDIA's single-chip approach keeps costs down.
So 3dfx's technology is probably pretty much dead. Not only that, but NVIDIA doesn't have any support liabilities as a result of this deal. 3dfx will be dissolved, and driver support for 3dfx products will evaporate. (So put that V5 on eBay now, kids.)
What about the rest of the graphics market? Via's S3 is in a holding pattern in the low end of the market. Matrox has pulled into its shell, living off the income it can extract from the G450. And newcomers like STMicro, with their Kyro chip, don't seem too threatening. That leaves NVIDIA and ATI.
Though some of you have said it ought to happen, ATI doesn't need to acquire anybody to get a hold of their core technology. The Radeon is slightly more advanced than the GeForce, so in terms of shipping products, they hold the graphics technology lead at presentbelieve it or not. Plus, ATI is trying to emulate NVIDIA's strategy, down to the part about having a new product every six months. (Remember, ATI already bought ArtX and has a north bridge chip with ArtX graphics integrated. They also have the Nintendo contract thanks to that acquisition. Shades of the Xbox deal.) I think they have the resources to do it, but it won't be easy to keep up with NVIDIA.
Odds are, ATI and NVIDIA will split the lion's share of the PC graphics market like Intel and AMD do in PC processors. The rest of the players will be left to pick up the scraps.