This is turning out to be a hardcore overclockers’ chip, because you have to do hardcore things to get a lot out of it. Provided that you’re not into voltage mods and heavy duty cooling, 1700 is unlikely. Even if you buy an 1800 or better, doesn’t look like
much more than 1700 on average with high-end air/1.85V.
These AGNGAs haven’t been showing up except with 1900+, and sometimes not even there.
For those less intense, wait.
According to XBit Labs, we’ll see the GF3MX being announced in December, and a new high-end GF3 in January.
nVidia cards have often been criticized for relatively poor image quality for items like text. Some have pointed out that cheap filter circuitry in the cards cause the problem and have suggested hardware hacks.
This review claims that Leadtek has put better filters in, and the results are as good or better than any card out there. We’re looking into this more, and will keep you posted.
Per visual quality, I’m beginning to suspect why most places have dropped this like a hot potato after trying this. I’m beginning to wonder if some of the respondents ever had to stay within a budget or meet a deadline.
I’ve come to the conclusion that testing can’t be designed by committee. Everybody wants something different, and everybody wants to add to the pile.
Just to give a simple (compared to what I’ve seen) example, if we put six video cards through seven tests using three monitors at three different resolutions at two color depths at three different quality settings, that’s 6 X 7 X 3 X 3 X 2 X 3 = 2,268 different tests. Assuming ten minutes a test, that’s two-and-a-half months worth of doing nothing but this.
And everytime a driver comes up, it’s another 378, or over a week’s worth of work? Sorry, but it’s not going to be that extensive. Nobody’s going to do something like that.
What I think we’re going to end up doing is waiting a week or two for a new card or two to show up and some beta drivers to stop being beta drivers, then start.