Well to say nV wasn't aware of DX9.0 specs is ridiculous, and a cop-out, the NV30 was pretty much mapped out before DX9.0 was finalized, heck the R9700 hit the market before DX9 was available in it's final form (still Beta), to think either had the chance to redesign with the amount of time required to make a chip, especially back then is ridiculous.
That statement also pretends that only nV and ATi were involved in the process; 3DLabs, intel and Matrox were all involved in this too. In fact both Matrox and 3DLabs lobbied to have some things added (displacement mapping) and removed (FP support for all stages). Wiki pretends that nV got shafted by someone else when it was in fact their own decisions that shafted them. Should 3DLabs blame ATi as well for the failure of it's P10?
Also the section "Rendering color support was limited to 24 bits floating point" is inaccurate, it wasn't limited unless you mean minimum is the lower 'limit'. nV picked FP32/15/FX12/int8 support for the NV30 (then crippled it for the NV35 to get more speed), but they still didn't have the ability to run in full FP32 in most situations at any reasonable speed, with a lack of register space and inefficient memory support.
How much influence do you think nV's lower support of specs on the GF4 series (PS1.3/DX8.0) versus the R8500 (PS1.4/DX8.1) had to do with that decision to not let them be outdone by anyone including 3DLabs' FP32 P10.
Also the R3xx series rendered everything in FP24, no FP16 no FX12, so even despite that extra work it was still a solid DX8.1 card because of the true 8 simple pixel pipes, which may not have been architecturally impressive, but did the job better becase the games weren't there yet to make a difference (heck they barely arrived by the time the X1900 did).
No mention of nV's 8x0 / 4x1 design decision which had just as much of an impact is a glaring omission, and something nV had said would make them better (only helped in heavy shadow / z-only situations [like D3]).
Also the mention of SLi i in the G7 series, when it came during the GF6 series (I remember questioning that 'extra connector' that nV kept calling and 'engineering port'. Xfire also came out with the X850XT and X700 series, just weeks before the X1800s were launched not after the X1K series launched.
The RSX is a modified 128bit version of the GF7900GT if anything, not the GF7800.
Interesting that there's no mention of the GF8800 delays either (much of which in reply to the Vista delay) as if it was right on time.
Current market share should be changed, nV is now #2.
Love the mention of this section;
"This has been responded to by some users who have made rude and derogatory posts in the NVIDIA forums"
'rude' customers, in a forum, no eh? :roll:
Also the issue was more than perceived, at the time nV forum mods in response to complaints about RTM support said that nV would have certified drivers by the time Vista was in retail (IIRC that was WaltC saying that), which didn't happen, they were still beta.
Also this part is of course incorrect; "NVIDIA does not manufacture video cards, only the GPU chips..."
They actually don't mfr the chips either, like ATi they are a fabless VPU maker.
Just another reason why Wiki is sketchy.
I doubt the ATi material, nor Matrox etc is any better. Maybe I'll check it at work.
(edit: typo R9700 out before DX9, not DX10 like I typo'ed)