ingtar33,
I'm really curious where you are getting your data from on this supposed Crossfire advantage because frankly I just can't find it. Yes, in a few isolated scenarios for games that are so new we don't see optimizations from NVidia yet, or in high resolution scenarios where the NVidia has less VRAM.. but on AVERAGE I can't.
This link is several months old but I'm having difficulty finding good comparisons that are newer:
http://www.tomshardware.com/reviews/crossfire-sli-scaling-bottleneck,3471-12.html
In particular:
"...but from a slight scaling advantage favoring the slower GeForce GTX 680s when you put two of them together."
*BTW, 2xCrossfire (HD7970's) only got 49% scaling on average with the above games using the Intel CPU and only 33% using an FX-8350 so your 90% scaling number seems too high. Maybe in a couple select games but not on average.
(I could give you a different link that shows closer to 80% scaling over the four games they chose so it totally depends on the games chosen. We really need a lot more games to average... but again I still can't find evidence that favours Crossfire overall.)
If you want to PM me with a link or two that definitively shows Crossfire winning I'm interested as I don't want to hijack this threat, but all my research shows:
a) Scaling is SIMILAR on average, and
b) SLI works smoother on average (see FCAT)
c) Scaling is far below 90% on average