Biggest ever 28nm gpu 70 types, 9 games, 3 resolutions

[joeyjojo]

Think this really shows how 980ti when it came out wiped out AMD's fury dreams. Once overclocked it destroys anything amd got.

You haven't got one have you by any chance?

 

[SouthEastBlue]

Unless I'm missing something, Nvidia seem to fair much better than people would have you believe in these benchmarks at least.

The average performance in all tests based around an AMD 7970 being at '100%' at 1080 and 1440 look decent for Nvidia too?

Spoiler

 

[Gregster]

Unless I'm missing something, Nvidia seem to fair much better than people would have you believe in these benchmarks at least.

The average performance in all tests based around an AMD 7970 being at '100%' at 1080 and 1440 look decent for Nvidia too?

Spoiler

You are correct and I found the same when comparing my Fury X to my Titan X and doing side by side comparisons. The Fury X was no slouch in fairness but the Titan X was a fair chunk faster. The 980Ti reference often gets put against the Fury X and this is where people gauge their results but truthfully, you look at OcUK's 980Ti selection and they are nearly all OC cards of some description and can add quite a few frames over a reference 980Ti.

What I do find odd about those results though is the 970 being faster overall than the Nano and I would have thought it was the other way round in truth but basing it over several bench results does pick out weaknesses and strengths of both AMD and NVidia.

 

[Rroff]

The most surprising results are those for the last generation cards - the AMD ones have aged far better in newer titles. A 290 is competitive with the 780 Ti! Looking back to the launch of each it was a very different story. AMD's architectures have usually been more forward looking and that's very much the case here.

Thing is though they are running a reference 780ti (876MHz base clock) against a reference 290 (947MHz).

In the realworld a "standard" 290 on sale would have been running 957-957MHz or so with some OC models at 1000MHz (a 5-6% clock speed increase over the reference model) while a "standard" 780ti would have been 1006-1020MHz (15-16% increase over the reference model) though what the actual boosts were would vary a bit making the exact performance increase over the reference model a little harder to pin down. So those results don't really reflect what people will be seeing in the real world.

(And ultimately Kepler seems to suck at The Witcher 3 which does weigh quite strongly into those results)

Its interesting that they have the 780 GHZ edition in there - while my card does boost on the higher side for those cards - its not so much above the average it would be making a huge difference to the numbers and in most of those games I see results which are much closer to the 970 with my out the box boost (though again this could be a little skewed without knowing the actual boost clocks their 970 and 780 GHZ are attaining - the 970 I have doesn't boost that great compared to what some people on here are seeing).

EDIT: Their 780GHZ v 780Ti results are a bit off as well - BF4 2560x1440 shows an 11% difference - if you look at the reviews for the GHZ edition even the ones that were on the lower end of the out the box boost range averaged 1.7% slower than a reference 780Ti at launch in BF4 at 2560x1440 and pretty much matches it over a range of games with less than 1% difference whereas they are showing an overall 5% lead to the Ti. Maybe they had a very good boosting 780ti and a poor boosting GHZ edition though the other results compared to the 780ti doesn't support that

EDIT2: Oh and my brother has just put a 970 in with his 4790K system which is a bit faster than my setup so gonna compare what he gets to my setup as well.

 

[Lokken86]

I think they got the clock speeds for the fury x and fury pro mixed up.