Originally Posted by Ninjawithagun
Should you decide to get the GTX980Ti instead of the Titan X (more so because of the cost difference), you will be happy to know that WItcher 3 can be played at max settings with a single GTX980Ti. I know because I own two GTX980Ti cards and don't have to enable SLI in order to get playable frame rates in Witcher 3! And yes, that's when playing at 3440 x 1440 resolution on my Acer X34 ;-)
I've done performance tests with a guy that owns 3x Titan X, this was GPU rendering in Iray. It turned out that one of his Titan X is approx. 46% faster than one of my older Titans, and that 4 of my GPU's actually performed slightly better than three of his Titan X's in iray, and we also did some game tests, with similar results, but with games favouring 3x Titan X more than 4x old titans.
There's no way I would choose a 980ti over a Titan X because I need the extra Vram for my work which is a priority compared to gaming.. plus the initial benchmarks done by TomH proved to be inaccurate, because they were using old drivers for the Titan X and old performance data, and new drivers for the 980ti that had been optimised for Maxwell, and other site like Anandtech, benchmarked using the same drivers and found the Titan X slightly faster, all be it by 2 fps. Of course now there are better alternative cooling solutions, so some benchmarks put the 980ti slightly above Titan X.
For me, doubling the vram is worth the extra cost, and a few extra fps more / less wont affect that decision, Makes a huge difference when Woking with large data sets, or complex scenes in 3ds max and iray,
If yo are on 980ti's in sli on a 3440 x 1440 monitor and still getting good fps in witcher 3 at max settings, then I am guessing its a scaling problem with SLI, as 4x older titans should be near par with 2x 980ti's, or a problem with my Mobo having to use SLI between two CPU's, but its never been a problem in the past, only with witch 3....
I definitely want to upgrade my GPU's though, pascal isn't that far away, 3 or 4 month IMO.