You may want to read through this article
which illustrates the pitfalls of microstuttering incurred with multi-GPU setups.
I have read that article, but did you look at the 6970 versus the 6970 crossfire numbers? This page
and this page
show that the crossfire setup has vastly superior frametimes in both cases. The SLI results are disappointing, but it looks like you saw the SLI results and concluded that multi-gpu setups are bad.
I would agree that microstuttering becomes less frequent as your graphics horsepower increases, but games are set to automatically detect and apply optimal video settings based on your setup...
If you roughly double the horsepower, the framerates at a given resolution/quality are going to be a nonissue because you're doubling the horsepower
. Who uses the default video settings? I kinda doubt the type of person who would buy a crossfire setup would.
Again, I would agree
that if someone is starting with no graphics card, then a single-gpu solution is really the way to go for various reasons. It's really semantics now that we know his mobo won't support it, but for many people out there the reality is that they already have one card, and to get the same performance that buying a second card would provide, they'd have to pay a lot more, oftentimes over twice as much.
I think Techreport's analysis of microstuttering is great. It just doesn't apply to all situations. To me, it's not a "waste" of power to have 2 6970s for a 1680x1050 display because I like eye candy and I like high fps. If you halve the refresh rate to 60, it would be complete overkill, but remember I'm effective running the same amount of pixels as a 2560x1440 monitor at 60 hz.
To me, the resolutions used in the triple-display gaming article are crazy because I don't care how much screen real estate I have if my framerates are in the 30s. But, I realize that to each their own. When you take the time to link articles, it might behoove you to make sure they support your point.