Power usage in your link is measured at the wall.
If they were getting 80% efficiency from the power supply, then 190W * 80% = 152W.
If they were getting 85% efficiency from the power supply, then 190W * 85% = 162W.
Fair enough on the benchmark. You found a game that scales to almost 90% gain from 2-way Crossfire.
From the GPU hierarchy, the HD 6870 is equivalent to the HD 7790 in a newer generation of cards:
http://www.tomshardware.com/reviews/gaming-graphics-card-review,3107-7.html
Here is a benchmark showing both the HD 7790 and GTX 780:
http://www.bit-tech.net/hardware/graphics/2013/10/08/amd-280x-270x-260x-reviews/4
HD 7790 - 39 FPS
GTX 780 - 111 FPS
Even if Crossfire could scale at 100%, this still isn't going to get 2-way Crossfire of HD 6870 cards anywhere near a GTX 780 or R9 290. 100% improvement on 39 FPS is 78 FPS. This is absolute best case scenario, certainly not typical.
That still wouldn't take into account the difference in video memory either (effective VRAM 1GB for HD 6870 cards in Crossfire compared to 3GB+ for the other cards you suggested), which is going to be a major problem at higher detail settings in modern games.
At the end of the day it is bad advice to compare Crossfire HD 6870 cards to an R9 290, GTX 780 or GTX 970.
You would be much better off with any of these cards than the crossfire setup.
Even an R9 280X or GTX 960 is going to beat the Crossfire setup every time.