Anyone looking to upgrade to the 980s should understand that with surround activated and running1080 resolutions that your performance could actually be less than even the older 680 card. The 980 is optimized for the higher resolutions and really doesn't give a big return until 4k resolutions are in play. There is no way to automatically setup monitor configurations when launching a particular game, so the process of establishing Surround, enabling SLI, or Maximizing 3D performance can be irritating and involved. SLI still cannot run more than two monitors, and if you wish to run three monitors (you could actually make it four) with an SLI profile you will have to run a third card with a different GPU.
If you wish to do this but all you have is a non-SLI motherboard you might try using DifferntSLI, which you can read about in many forums across the web. I don't have actual experience with that.
In order to get this setup into full swing I have settled upon using a 4K GSync monitor as the center monitor of a surround setup. One added benefit of getting a GSYnc monitor is the locked refresh, and the one I bought (an Acer 28") does 1080p gracefully, and restores to 3840 x 2160 without complaint also. That way I can use SLI and DSR at 4K and then step back to surround for working on the same system. There simply is not enough VRAM to use three monitors at 4k, sadly. SLI is active when using surround, but the unique SLI anti-aliasing features are not available until the "Maximize 3D Performance" option is selected, which means a single monitor.
I tested 2-way SLI on a Z87 MB, and 2-way, 3-way, and Quad SLI on an X79 MB. The Z87 system uses some additional devices in the available PCIex1 slots and the final PCIex8 slot without affecting the available GPU lanes (GPU-Z reports both cards running at x8). The X79 system is capable of running two cards at x16, or four cards (Quad SLI) at x8. As we have discussed before, there is no real advantage to running a single card at x16 vs x8, however what Quad SLI offers is the ability to use even higher AA settings. This should be further improved once the drivers are updated to add MFAA functionality.
Testing AH offline is not an optimal method of benchmarking, I know. Offline with every graphics option active I ran a few tests in tanks at a large field, with the B29 zoomed in to view the environment effects, and with the Arado right over the top of the city strats and compared it to duplicate tests on a single 680 GPU running Surround. The Arado test on the Z87 system saw a loss of performance at 1080 980 surround (134.7 fps)compared to the 680 Surround (139 fps). It's only a 4fps loss, and I mention this simply to point out that standard resolutions do not require this setup. The same setup with the B29 Benchmark saw an advantage return to the 980 surround setup as frames increased by more than 30fps. Please, bear in mind this is not an optimal testing method and there are simply too many variables to get consistent results.
Looking forward to more from Gman and the 1440p results.