2x780TI Gigabyte Ghz Edition Valley benchmark: imgur.com/a/tyJIx#1 
2x970 Reference design: imgur.com/a/tyJIx#4
2x980 Reference design: imgur.com/a/tyJIx#5
2x970 Asus Strix cooler (custom pcb and cooler) imgur.com/a/tyJIx#7
2x980 Asus Strix cooler (custom pcb and cooler) imgur.com/a/tyJIx#8 keep in mind for this one, this is on the x99 platform that has lower clock speeds, so the comparison to the 2x970 isnt fantastically scientific, I am just using the systems I can, while I can.
 
i.imgur.com/KIVrb0I.jpg - Forum member mcneill467 2x980 sli setup As you can see these are considerably higher than what my results with the 980s are getting so I will update the driver version Im using then update this post further.
 
2x980 Evga Super Clocked Editions: i.imgur.com/TViLlUJ.jpg
 
 
I want to start this thread off by saying, I know that A whole bunch of you know this already, I am making this post because I really want to try and help those guys who are not in the business and don't exactly know what the sweet spot is for price to performance.
 
***Quick note*** I know because I am mixing vendor's this isnt exactly lab conditions of testing, but im trying to give examples with the information I have, and the resources I have available to me.
 
Okay, so with graphics cards there is a consumer sweet spot for price to performance, sometimes you can really want to just shove money at a product to say you have the flagship one, honestly the 970/980 is not the place to do this at, I really mean it, not at all.
 
I have a variety of different benchmarks for you, some of 970 sli, some of 980 sli, different coolers, different vendors, please just look at the score the graphics cards get, the min and max frame rates.
 
Some times the 970's in SLI even beat some of the stock 980's SLI...
 
Here is the link to the Imgur folder: imgur.com/a/tyJIx#0
 
I would like to draw your attention to just a few of the main ones:
 
Let's start of basic, let's go back a generation and look at a good 780ti SLI setup, it's the gigabyte ghz editions, high clock speeds, good card all around: imgur.com/a/tyJIx#1 Valley score was 5472, and I will mention it here, this is the HIGHEST score of any of the cards I have in the selection thus far. (now it was running on an overclocked 4790k, running at 4.6) That really shouldnt affect valley scores above about 100 though.
 
So why proceed to the newer cards if the 780ti beats out all the competition? Well, the drivers are new, optimizations need to be made, the 780ti were pretty much the end line 780ti's the max performance was pulled out of them, so it's unlikley we will see any better 780ti's, but we can see the 980's improve with a TI version.
 
You also may be interested in the 900 series cards for the same reason I am, lower noise, lower temps, and lower power. I dont care too much about performance, I only play at 1080, so whatever high end card I get is going to breeze through that.
 
So next in line let's look at the reference 970's: imgur.com/a/tyJIx#4 and the reference 980's: imgur.com/a/tyJIx#5 Okay so as you can see the 970's got a score of 4341, more than 1000 less than the 780 ti's do. The 980's get 4856, so if you look at score wise on valley, it's about 500 more point's for 2 980's than 2 970's, however the price difference you're looking at about £400/$600, now, im not saying valley scores are everything, but they are a good indication of what the cards will perform like in real life gaming use. 
 
If you just take a look at the min/max frame rates of both of those cards, they are pretty much identical.
 
Okay, taking a step away from the reference designs and looking at some benchmarks of the aftermarket coolers and the power houses they should be... Right? No...
 
Keeping it a little more consistent though the 980's get a better/worse CPU depending on how you look at it, our 2 980's actually had to go in an x99 build so you will have to forgive that, (lower clock speed, more cores, should generally equal out for performance) Okay, so the 970's were the Asus Strix design: imgur.com/a/tyJIx#7 as you can see they get a score of 4525, fantastic that's about 200 more than the reference designs. The 980's are the same Asus Strix, however... imgur.com/a/tyJIx#8 they get a miserable 4750, this is less than our reference design's from Gigabyte.
 
Okay now what you're probably into most seen as you're on the EVGA forum, some EVGA SC card result's and the scores, unfortunately I was only able to get 2 of the 980's SC Acx 2.0: i.imgur.com/TViLlUJ.jpg Yes that's right 2 980's got 4099 in Valley, this is the lowest I have ever seen 2 980's get.
 
I love EVGA and they have my brand loyalty, I just really dont know what to say other than, what happened with the 980's? Are you guy's noticing they are not performing all that well? Is it purely software based or are the optimizations of Maxwell cutting corners leading to leakier chips?
 
This post wasnt to dig the evga scores though, it was to show you why you shouldnt buy a 980, or even 2 980's.
 
Conclusion. The Nvidia GTX 980 is a fantastic flagship graphics card, but not for it's price tag. If you are only in it for the performance, you can get basically 2 970's for about £100/$170 more than 1 980. 
 
P.s EVGA Team, if you wan't to send me any 970's to test I would happily oblige. Finding it very difficult to get 2x970's (I even want 2x970 ACX 2.0 in my personal rig, just cant source them) Or even if you're not up for sending me any, I would really appreciate anyone at EVGA or the Community popping 2 of the 970's in a rig, let valley idle for 1 hour (that is what the case was for these images) and then benchmark it and send me the image.
 
P.P.S, waiting for all of the TLDR comments :D
 
 

People that are on the fence about whether to get a 980, should consider the 970. Its a great solid card and amazing price/performance ratio.
 
However, for those of us who have already made up our minds about getting the 980, there is no amount of convincing that will change that

You make a good and valid point, everything in hardware is subjective to your position and what your needs are. I really wish my link's worked so you could see the benchmark images :3

You can post a link as long as you leave out part of it. (For members under the threshold) 
 
Like if posting a newegg link, leave off the httppwww...you know. Just do newegg.com/product cuz technically it isn't a completed link.
 
At least, that is what I did

LiMiTeD_DiAvOlO
Another thing to keep in mind, you are acknowledging that the 780ti's are at the end of there life cycle, performance update wise. Any improvement at this point will be nominal pending some breakthrough software tech.
So to be making these accusations right out the gate that the price to performance ratio is bad based on Valley results, doesnt mean much. The price to performance ratio could easily get higher given the improvements some optimization could provide.

Your comment about 2 970's running in SLI outperforming a single 980 is not surprising at all and shouldnt be shocking news to anyone. You will get better performance but at a higher price than one 980, which was a bad ratio to begin with according to you...not really sure what you were trying to tell people, not to spend the money on one 980 but more money on 2 970's instead? If they only cared about performance that is. If they only cared about performance the cost was never an issue and they will be running 2 or 3way 980cls.

TL:DR
Way to early to compare a 980 to its dual chip predecessor that has had many updates

LOL no you misread, 2 970's outperformed 2 evga's 980's... That is shocking 

I am really doubting 2 970s beat 2 980s. We have seen numerous sites such as hitechlegion and [H] show that sli980s beat sli 970s.

My 2 GTX 970 SC ACX 2.0s are getting 5160 @ 2560X1440. On a measly I7 4790K @ 4.6Ghz.... Not a particularly blazing system.

LiMiTeD_DiAvOlO

jpcopeland1
My 2 GTX 970 SC ACX 2.0s are getting 5160 @ 2560X1440. On a measly I7 4970K @ 4.5Ghz.... Not a particularly blazing system.

on a measly CPU that doesn't exist.  MEASLY I tell you MEASLY!!!
 
I retract my statement kinda.  After looking at your sig you either meant 4790k, or im drawing a blank on the 4970k.  been known to happen...those blanks

Think he means 4790k, there is no 4970k. I wouldnt call that measly considering that is better than most systems.

bowler484
There are too many GTX 980 SLI reviews out there to believe what the OP is saying word for word. 
 
I can see an high clocked aftermarket 970 touching a 980 performance though.  But this has happened almost every time in the past as well.  670 could OC to 680.  780 could OC to 780 Ti etc.  Nothing has changed there.   
 
The indisputable fact is that the 980, when comparing reference apples to reference apples, is over 10% faster. 
 
Is that 10% worth the premium.  Most of the time no, especially at 1080p.  But at 4K like myself, that 10% can mean getting a 60fps average or not getting a 60fps average.  That means the 980 SLI would have a longer lifespan in my rig making the premium worth it for me.
 
So making a blanket statement that one is better than the other is quite silly.  Everyone's needs are different.       

Preach on 4K brother.

seta8967

LiMiTeD_DiAvOlO

jpcopeland1
My 2 GTX 970 SC ACX 2.0s are getting 5160 @ 2560X1440. On a measly I7 4970K @ 4.5Ghz.... Not a particularly blazing system.

on a measly CPU that doesn't exist.  MEASLY I tell you MEASLY!!!
 
I retract my statement kinda.  After looking at your sig you either meant 4790k, or im drawing a blank on the 4970k.  been known to happen...those blanks

Think he means 4790k, there is no 4970k. I wouldnt call that measly considering that is better than most systems.

But not as good as the 5000K series CPUs. Come on tax time.

LiMiTeD_DiAvOlO
Another thing to keep in mind, you are acknowledging that the 780ti's are at the end of there life cycle, performance update wise. Any improvement at this point will be nominal pending some breakthrough software tech.
So to be making these accusations right out the gate that the price to performance ratio is bad based on Valley results, doesnt mean much. The price to performance ratio could easily get higher given the improvements some optimization could provide.

Your comment about 2 970's running in SLI outperforming a single 980 is not surprising at all and shouldnt be shocking news to anyone. You will get better performance but at a higher price than one 980, which was a bad ratio to begin with according to you...not really sure what you were trying to tell people, not to spend the money on one 980 but more money on 2 970's instead? If they only cared about performance that is. If they only cared about performance the cost was never an issue and they will be running 2 or 3way 980cls.

TL:DR
Way to early to compare a 980 to its dual chip predecessor that has had many updates

 
I never actually said that it was surprising that 2 970's beat 1 980, I actually had an indicent where the 2x970's beat out 2x980's which is surprising and should make everyone wonder just what is going on with these new chips.
 
I agree, the drivers and constant support nvidia gives to there gpu's will see an increase in performance for both cards, but at this point in time, we dont have that, we can only go off what we have right now.

Id be interested to see 2x970s pushing near 20k firestrike.

I have also updated the links, as a lot of you decided "you cant believe that the difference is so little" Well luckily valley tells you all the information about the system, and if you dont actually think the gpu is being utilized, in valley at the top right it tells you the temps, there is no reason a gpu would go about 50c if it wasnt being utilized.
 
Now as I mentioned in the OP I know this isnt scientific fact but it is a good indication of how these cards are performing and what the difference in performance really is.

I can see in your footer, you have 2x980's so I can understand that you wouldnt want your purchase to be going unjustified, if it's justified for you, that's all that matters man. The 980 is a fantastic card, I dont doubt that, it just isnt pretty when you consider the price premium you pay for the little performance. As I only have access to these machines for a little amount of time before they have to be sent off, I cant do other benchmarks, we run valley as a testing program to make sure the GPU's are performing well and not artifacting. Please if you think that I have just been unlucky with my graphics cards then run valley (let it idle for 1 hour) then run the benchmark, those were the conditions of all these tests, though if you dont have the hour, I would still like to see your valley screenshot of the benchmark. 
 
Reply with an imgur link and I will add it to the OP.
 

mcneill467
I used "step-up" so the price premium was not a factor.

 
Ah that's cool :D I would still be very interested in your valley benchmark screenshot if you would be willing to run it. The more systems we have in this plethera of result's the better indication of the performance, + if some cards are just not as strong as others.