Just musing a bit here. Someone please let me know if I've got my assuptions mixed up, since I've been away on vacation for almost 2 weeks and haven't been able to keep up with every post...

Pat has mentioned that at the annual meeting, he specifically asked Ballard if OEMs were sampling silicon of the next-gen product for their fall refresh, and was told "Yes".

But last week, 3dfx announces the T-buffer technology, and claims that it will be included in all future products. But to demo it, they do an emulation on an 8 chip V2 Obsidian board.

This would seem to imply that they do not yet have any working silicon that incorporates the T-buffer in hardware. So what exactly are the OEMs testing right now? The above would indicate that the true next gen part is not yet available (or why not use it to actually demonstrate the technology they were trying to showcase last week)?

What have the OEMs got in their hands, a V3 'Ultra'?, i.e. a V3 chip with 32 bit color, true AGP4x, 32 Meg DRAM, running at, say, 200 MHz? These were the bullet items missing from the V3 initially, and I can't imagine 3dfx trying to offer anything for the fall without them. Would adding these items have been a worthwile 'step', or a waste of effort, to produce yet another stopgap solution for the Fall OEM cycle?

This would allow them to offer a lower price solution to the OEMs to use as the standard video card (where the true volume is). The true next-gen part (V4?) would then be released to compete with the NV10 in the high end.

If 3dfx doesn't have a true next-gen part sampling with OEMs right now, then the above speculation (as distasteful as it is) BETTER be true, because it seems as if NVIDIA does have at least alpha silicon of the NV10 available (if Tom can be believed in any manner what-so-ever), so the NV10 has a good chance of coming out well before the V4.

I hate even speculating this way. Is it possible the 3dfx does have V4 in silicon, but chose not to use it for the T-buffer demo because of software issues? This seems absurd to me, but is it likely?

Wavelength