There's something about 3GB against higher memory capacities i have to mention.
You see, benchmarks are being done in what i call "offline-scenarios". It's synthetic benchmarks and demos being done on rather lower demanding enviorments.
While review figures can show no differance between 3GB to 6GB for example, real-life scenarios could show it very well to some users.
I'll take myself as an example of a user that have a 1440P monitor with a single GTX 780 Ti 3GB. I play a lor of multiplayer games and some of them have very large enviorments like ARMA III or Battlefield 4.
Monitoring GPU memory usage, i mostly scrape the 3GB barrel at the highest settings. Usually using between 2700 and 3000MB in those games.
Now, the problem is testing this. With offline tests you can have full control over the tests. Everything is 100% the same each time you test for maximum reliability in results, unlike multiplayer situations that change in every test.
It is worth noting that some games today use very-high resolution texture mixed with a lot of render distance that just hog on GPU mem. If you will put my Ghz Edition GTX 780 Ti or this STRIX in-front of my, i might as-well pick the STRIX.
