Originally Posted by Somasonic
I don't know what the comparative performance of these cards should be but from what you've said we could possibly conclude that your results are about as expected.
If 980 Ti = 780 Ti +10%
we could suppose that 980 Ti SLI = 780 Ti SLI + 10%
We know that that the 1080 Ti is roughly equivalent to 980 Ti SLI so we can then say
1080 Ti = 780 Ti SLI +10%
Which would be about in line with what you're seeing.
This could all be a steaming pile in the real world so I'm happy to be corrected
I've seen so many benchmarks in the past years that I might have hyped my expectations.
GTX 1070 is 3% faster than GTX 980Ti
GTX 1080 is 23% faster than GTX 980Ti
I remember not picking up 980 Ti because it was in many scenarios that 5-20% performance increase. Worst case a few fps, best case a noticeable difference, and of course the DX12 marketing.
examples firestrike http://www.overclock.net/t/1406832/single-gpu-fire-strike-top-30
and firestrike extreme http://www.overclock.net/t/1443196/fire-strike-extreme-top-30
Looking at firestrike scores in the benchmark sections I see to a 20-25%+ improvement on the 980Ti.
Then another 20-25% for 1080 to 980Ti
Last 20-25% for Titan XP to 1080
The extreme scores are is very similar
I'm trying to understand how in this large gap of architectural improvements over 2 generations that a SLI 780 Ti is putting up the scores close to this 1080 Ti. Could it be another 50mhz will make 5-10% out of this card? Something just doesn't seem right with my numbers, maybe the custom 780Ti's are very strong and still capable or the reference 1080 Ti design is really held back with it's power draw?