DemoCoder said:
I already said that the performance argument wasn't purely FUD (it's technically true). If ATI strictly stuck to the facts of PCI-E vs AGP performance, I wouldn't call it FUD. (I might call it something else) The FUD part is speculating about incompatibilities and defects. There is no proof of these two, and it's something that *may or may not* be true. It's not certain, but it's trying to cast doubt, uncertainty, and fear. DUF!
Let's be honest, it's not native PCIe support, it's not the equivalent, either. It's a nVidia-only, custom implementation of "AGP x16," and nVidia calls it exactly that, from what I've read. I don't think it's important--it's only temporary--nVidia won't be using it long, I'm sure. It's just an interim step and not worth defending on that basis alone, seems to me. Depending on how R4x0 stacks against nV40, it may or may not even be even a temporary issue worth much discussion. I think we've given it much more than it deserves already...
The future is PCIe. I don't think that's in doubt. As far as what ATi is speculating about, if it was reversed and nVidia was doing native PCIe with ATi doing its own custom "AGP x16" instead, I have no doubt nVidia would be making similar comments, and so nothing ATi has said about it thus far (very little) should be surprising, really.
There are two kinds of FUD, in my opinion, constructive and destructive. For instance, telling people that it isn't wise to overvolt and overclock .09 micron cpus to any great degree because you might well destroy the cpu much more rapidly than otherwise, because of electromigration and other things, is certainly spreading Fear, Uncertainty, and Doubt about overvolting/clocking them--but because the FUD is based on fact it is actually constructive and helpful to people who might not already understand the issues involved.
As you point out, whether or not nVidia's custom overlay of "AGP x16" on top of motherboard circuitry expecting PICex16 will ever be problematic is something that "may or may not be true" and is presently unproven either way. Therefore, I don't think it can be FUD, but is rather, at worst, mere speculation.
An example of destructive FUD, I think, can be found in nVidia's various ramblings of a year ago concerning how "ps 2.0 wasn't the future of 3d," and of how 1.x, instead, certainly was...
Since ATi's own R3x0 architecture convincingly disproved nVidia's PR hypothesis concerning the "inappropriate nature" of ps2.x, and indeed their recent nV40 also convincingly disproves it, I think we'd be safe in calling those comments destructive FUD as opposed to unproven speculation. FUD which cannot be proven, most especially FUD which can plainly be proven false, is really what characterizes "destructive" FUD, IMO.
Should it be shown in the future that nVidia's AGP x16 operating in a PCIex16 environment causes no problems (regardless of any performance differences that may be shown), and should ATi continue to state that it thinks there is a possibility that it might cause problems for PCIex16 environments even after it becomes known otherwise, then that would certainly be FUD of a destructive kind. But at this stage of the game I don't think that FUD is as accurate a description of those comments as the phrases "reasonable speculation" or "prudent speculation" might be at present.
