GTX 980 Ti Strix OC vs GTX 1070 Strix OC benchmarks

[Artas1984]

Greetings once more!

In recent times i had the chance to test Asus GeForce GTX980 Ti Strix OC Gaming and Asus GeForce GTX1070 Dual video cards in this thread. GTX980 Ti outperformed GTX1070 by 6 %, and did it in every one of 20 tested games! Granted, Asus GTX1070 Dual featured a very small 38 MHz boost overclock from reference 1683 MHz to 1721 MHz, while Asus GTX980 Ti Strix OC Gaming featured a massive 241 MHz boost overclock from reference 1076 MHz to 1317 MHz - having this in mind i expected the Strix model to come on top, despite having in mind that stock GTX1070 is faster than stock GTX980 Ti. However, i felt like i did not do justice by comparing a high clocked GTX980 Ti vs. a low clocked GTX1070, therefore i bring this battle back to fair grounds, since i've got the equaly branded GTX1070 Strix OC Gaming model, which features a hefty 177 MHz boost overclock.
Asus GeForce GTX980 Ti Strix OC Gaming VS. Asus GeForce GTX1070 Strix OC Gaming

BENCHMARK HARDWARE & DRIVERS

Intel Core i7 5775C 4.3 GHz
Gigabyte Z97X-Gaming 3
Crucial Ballistix Tactical 2X8 GB DDR3 1600 MHz C8
Samsung Evo 500 GB SSD (all games are installed in it)
Windows 7 Pro X64 (updates installed)
NVIDIA Forceware 383.33

For this test i will be using 2560x1440 resolution, all settings max, and custom anti-aliasing. Since the last benchmark i've updated NVIDIA drivers from 378 to 383, overclocked CPU from 4 GHz to 4.3 GHz and included some new games, excluded some older ones.

Let's begin!

ALIEN ISOLATION (SMAA 1TX)

GTX980 Ti slightly pulls ahead. These scores are constant within many repeats.

ASSASSIN'S CREED SYNDICATE (MSAA 2X + FXAA)

GTX980 Ti is a bit faster and the frame rates are pretty constant within many repeats.

BATTLEFIELD 1 (NO AA)

GTX980 Ti is only pulling ahead by a mere thread and the frames are more less constant within many repeats.

CALL OF DUTY BLACK OPS 3 (SMAA 1X)

From the many repeats GTX980 Ti scored the best average score, which is significantly better than that of GTX1070. The scores are very inconstant and the test needs a lot of runs to be legit.

CRYSIS 3 (MSAA 2X)

As far as i am concerned, this is still the best looking game i've ever seen. Crysis 3 is also the second most laggy game ever seen after Watch Dogs 2, but GTX980 Ti was constantly slightly better than GTX1070 in all tests.

DEUS EX MANKIND DIVIDED (MSAA 2X)

The frames feel very stable within the many tests and GTX980 Ti is ever so faster.

DRAGON AGE INQUISITION (MSAA 4X + PP AA HIGH)

Another stable FPS game with a tie between the cards.

DYING LIGHT (NO AA)

Very small advantage for GTX980 Ti. Stable FPS.

FAR CRY 4 (MSAA 4X)

Another tie between the cards. This test needs many tries as the FPS can vary quite a bit within the runs,

MAD MAX FURY ROAD (NO AA)

GTX980 Ti now smashes GTX1070 to a surprising result.

MASS EFFECT ANDROMEDA (FXAA)

Another tie between the cards. This game can be very unstable and the FPS might drop and rise within the same scene. However this specific 15 second dialog scene runs pretty stable every time benchmarked.

METRO LAST LIGHT REDUX (SSAA 2X)

Super sample AA was used for future SLI GTX980 Ti tests in order to provide efficiency. That being said, GTX980 Ti is just barely faster than GTX1070. The FPS feel constant in this game. Not using the in-game build benchmark, because it sucks.

MIDDLE EARTH SHADOW OF MORDOR (NO AA)

That's the first and last in-game build benchmark used in this video card battle. It is reliable, since the results don't vary so much from test to test.

MIRROR'S EDGE CATALYST (AA post 0 - unknown)

I don't like games which have no option to change AA in the menu, this is the main reason i will never use Fallout 4 in any of my benchmarks. Having said that, just like in Battlefield 1 GTX980 Ti has a similar advantage over GTX1070.

PREY (SMAA 1X)

Not to be confused with the 2005 game Prey which ran on Doom 3 engine. In this game GTX1070 is getting a real beat up.

PROJECT CARS (AA DS6X + SMAA ULTRA)

Interesting AA modes, selected the best one in order to keep FPS smoother and keep the game still playable.

RISE OF TOMB RAIDER (SSAA 2X)

Used super sample for future SLI optimizations. The official benchmark provides unstable results each time being run, so me getting tired of it simply ran the start of the game sequence for 15 seconds. GTX980 Ti is about 10 % faster, which is HUGE for a card, that is supposed to be an equal competitor.

SNIPER ELITE 4 (NO AA)

The newest game in the bench list feels great. Clear advantage for GTX980 Ti.

WATCH DOGS 2 (PP SMAA)

The most laggy game i've ever seen. The FPS can unexpectedly rise and drop in this simple start scene. Calculating the results of some 6 runs for both cards, GTX980 Ti was always ahead of GTX1070 by a little.

WITCHER 3 WILD HUNT (NO AA)

Each time tested, GTX980 Ti would always beat GTX1070 by the very minimum of 1 FPS.

-------------------------------------------------------------------------------------------------

AVERAGE FPS PERFORMANCE SUMMARY

So, i have compared the performance of two different video cards in the same Asus Strix OC Gaming brand and as far as things go GTX980 Ti Strix OC model is still superior to GTX1070 Strix OC model by some 4 %. Sure, you can not compare these results with my last benchmark results because of different games, different AA settings, different testing places and new drivers.

I was disappointed with GTX1070, but what ever! Take comfort in the fact that GTX1070 requires much less power and produces less heat. Also, only DX11 games were tested, so this is not a sealed win for GTX980 Ti either. Their noise level is equally awesome from both of these cards. However, performance is the most that matters to me!

Next time (50 % chance of happening): GTX1080 Ti vs. SLI GTX980 Ti.

 

[cdawall]

So just stock no overclocking on both....

 

[peche]

would like to see GTX 980 non ti numbers....

 

[EarthDog]

Thanks.

May I ask why you use very low AA settings in (most) of these games? Especially the ones with sky high fps...

Wondering of using higher settings it would matter. The 980ti was always a brute, but wondering why your results seem to differ...tpus testing showed reference to reference at 2560x1440 is 12% in favor of the 1070. Many sites are in that ballpark, in fact.
https://www.techpowerup.com/reviews/MSI/GTX_1070_Gaming_Z/

would like to see GTX 980 non ti numbers....

look at the tpu review... you can see these results there too.

 

[Artas1984]

So just stock no overclocking on both....

Point taken! I CLEARLY get your frustration that GTX1070 lost! I will overclock both of the cards to Techpowerup specs, and provide only the final performance summary! I expect to do this in the weekend, so follow the thread! I am risking of voiding my warranty by doing this as i will be adding milivolts to both of the cards. For now i just wanted to get the basic factory results as fast as i could. I already wasted 3 months till i got the needed GTX1070.

Thanks.

May I ask why you use very low AA settings in these games?

Yes. As you earthdog, and all of You have noticed, in some games i do not use AA at all. That is because of two reasons:

1. There is no explanation what type of AA is being used (for example Dying Light) - i hate that. That is so lame, shame on developers.

2. SLI scaling is horrible when using those "unknown" POS AA modes. I only use clearly defined ones - SS, MS, SM, FX, and for some reason SLI scaling is good with them, but whenever i use some unknown AA, SLI scaling is bad - therefore i will not use those unknown AA modes at all.

 

[EarthDog]

I nija edited you.. please check the above post again.

1. Thumbsup
2. But whwn you used 'named' ones, you used 1x/2x in most cases. You can easily see jaggies there at that res and in many of the titles the fps are plenty to use at least 4x or even more.

Just wondering how they respond when a more typical use for aa is used...and why your results difdere so much from others.

Last, what were the ACTUAL boost clocks in testing? What gpuz says is the monimum boost, when in reality, these both boost much higher.

 

[DRDNA]

Nicely done and presented cleanly!

 

[cdawall]

Point taken! I CLEARLY get your frustration that GTX1070 lost! I will overclock both of the cards to Techpowerup specs, and provide only the final performance summary! I expect to do this in the weekend, so follow the thread! I am risking of voiding my warranty by doing this as i will be adding milivolts to both of the cards. For now i just wanted to get the basic factory results as fast as i could. I already wasted 3 months till i got the needed GTX1070.

I actually have one of those strix 980Ti's in a mining rig running 1480 on the core lol I am just curious what the pair you have hits. I only have the generic low bin 1070's to test with.

 

[Artas1984]

I nija edited you.. please check the above post again.

1. Thumbsup
2. But whwn you used 'named' ones, you used 1x/2x in most cases. You can easily see jaggies there at that res and in many of the titles the fps are plenty to use at least 4x or even more.

Just wondering how they respond when a more typical use for aa is used.

Aaa. Right. I could have used 2X AA instead of 1X AA in those instances for Prey, Call of Duty and Alien (they have no 4X AA mode). It is just that i did the tests for SLI GTX980 Ti a month ago, and i selected those 1X AA modes for best SLI scaling, and since i had the results, i wanted to keep them the same here in case i will have gotten SLI GTX1070 in the future past.
When it comes to 2X AA in some of the games used - it is ok, earthdog, they are MS or SS based and enabling 4X AA stomps the performance of those games so bad, that the games are simply unplayable with GTX980 Ti or GTX1070 at 2560x1440 and max settings. Watch Dogs 2 is unplayable even with 2X MSAA...

 

[EarthDog]

and i selected those 1X AA modesk9ving for best SLI scaling and

not sure cherry picking for best results is a good idea...

Im only saying to use enough AA in titles where fps will be ok to do so...and if that has anything to do with your results vs others on the web.

Last, what were the ACTUAL boost clocks in testing? What gpuz says is the minimum boost, when in reality, these both boost much higher.

 

[Outback Bronze]

Nice review!

Id still pick the 1070 any day of the week.

Pretty hard to find a 980Ti brand new with warranty.

I'm not sure if a 4k review would have been a better representation to see if the 6GB vs 8GB comes into play, but I suppose adding 4-8x AA can do the same job. Also I don't think these are 4K cards.

Still very nice review mate. N1

 

[Artas1984]

Last, what were the ACTUAL boost clocks in testing? What gpuz says is the monimum boost, when in reality, these both boost much higher.

GTX980 Ti: 1418 MHz
GTX1070: 2025 MHz

 

[EarthDog]

Any thoughts on why these results are different than others on the web?

 

[Folterknecht]

Any thoughts on why these results are different than others on the web?

I think most of the time, you only see Ref 980Ti in the comparison charts. Especially against custom 1070s with better cooling (no thermal throtteling) and OC out of the box, that results in misleading charts.

 

[The Pack]

If you have some Samsung Rams on your 1070GTX and make a litlle bit more OC with these Rams, the 980ti has no chance in all games and Benches...

 

[EdInk]

Considering the 1070 replaces the 970, not 980Ti, I think it performed well

I'd take the 1070 for its larger VRAM capacity and efficiency any day, if money wasn't a barrier.

Interesting results. Thanks pal

 

[P4-630]

I will overclock both of the cards to Techpowerup specs, and provide only the final performance summary!

Nice, waiting for the results!!

 

[The Pack]

Considering the 1070 replaces the 970, not 980Ti, I think it performed well

I'd take the 1070 for its larger VRAM capacity and efficiency any day, if money wasn't a barrier.

Interesting results. Thanks pal

Yes it is i think, but its more than efficiency like a 980ti, the Pascale GPUs are very good in Power/Price. In Switzerland it has Cards between 300 and 400 Dollars. It isn`t to much. At the end of the year, you must look, what you have.

 

[EarthDog]

I think most of the time, you only see Ref 980Ti in the comparison charts. Especially against custom 1070s with better cooling (no thermal throtteling) and OC out of the box, that results in misleading charts.

in my link, tpus review, it was ref vs ref with the 1070 coming out 12% ahead at 25x1440. The overclocked card is a few more percent faster than the ref 1070. These are similarly overclocked and it comes out 4% behind. A difference of 16%.

 

[peche]

look at the tpu review... you can see these results there too.

i mean against GTX 10XX series

 

[EarthDog]

i mean against GTX 10XX series

look at the tpu review... you can see these results there too.

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080_Ti/30.html

 

[peche]

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080_Ti/30.html

you know, sometimes i could be way to lazy

have a beer mate

 

[EarthDog]

Its a problem on the internet, and partially why I dispise forums... the lazy gather here and just beg for free answers (you do not typically do that).

 

[peche]

Its a problem on the internet, and partially why I dispise forums... the lazy gather here and just beg for free answers (you do not typically do that).

i was searching some information about the card, decided some days to move on into GTX 10XX series, was saving some pennies, and i was thinking about getting GTX 1070 for a decent price, then realized that im not into computers that hard, also noticed that GTX 980 still have plenty potential... but you know seldom i think if i should or not get the new video card, also some days the question of upgrading my whole rig comes just to ink my day in black ...

Regards,

 

[Folterknecht]

in my link, tpus review, it was ref vs ref with the 1070 coming out 12% ahead at 25x1440. The overclocked card is a few more percent faster than the ref 1070. These are similarly overclocked and it comes out 4% behind. A difference of 16%.

What do you think, which card is more hindered (performance wise) by the refence cooler and stock power target - 980Ti or 1070?