• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

videocardzFirst NVIDIA GeForce GTX 1080 3DMark Benchmarks

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Okay, I thought I would wait until the Big Pascal and Big Vega drop next year to replace my 2 R9 290s in rig 2 below.

I don't mine with the 290s, just game and both rigs are custom water cooled so I have to buy a new gpu block no matter what card I go with.

RS and I have been discussing options and an option came up about selling the 2 R9 290s, moving the GTX980 TI to replace them and replace the 980 TI with a 1080 in the 5960x rig (no need to "upgrade" to 6950x).

Your thoughts?
 
Okay, I thought I would wait until the Big Pascal and Big Vega drop next year to replace my 2 R9 290s in rig 2 below.

I don't mine with the 290s, just game and both rigs are custom water cooled so I have to buy a new gpu block no matter what card I go with.

RS and I have been discussing options and an option came up about selling the 2 R9 290s, moving the GTX980 TI to replace them and replace the 980 TI with a 1080 in the 5960x rig (no need to "upgrade" to 6950x).

Your thoughts?

sounds like a plan🙂.but how fast is a 1080 going to be?hopefully blistering fast but a meaningful upgrade from a 980ti? hmmn not sure time will tell.
 
Okay, I thought I would wait until the Big Pascal and Big Vega drop next year to replace my 2 R9 290s in rig 2 below.

I don't mine with the 290s, just game and both rigs are custom water cooled so I have to buy a new gpu block no matter what card I go with.

RS and I have been discussing options and an option came up about selling the 2 R9 290s, moving the GTX980 TI to replace them and replace the 980 TI with a 1080 in the 5960x rig (no need to "upgrade" to 6950x).

Your thoughts?

I would wait for Polaris and Pascal to have a better knowledge of the GPU performance and perf/$ in Summer 2016, with more DX-12 games released until August (WarHummer and Deus Ex).

You could also replace those 290s with 2x Polaris 10 or wait for Vega 😉
 
Mm...it has the performance of a R9 295X2 on a single card. It's interesting. Because it will bring down all 28 nm cards.

BTW...Intel HD is doomed again..if a mid-high class chip can do that, wondering how a low tier card can do.

If they don't rebrand any desktop cards and are all in the new uArch... Intel HD would be outclassed hard.
 
Mm...it has the performance of a R9 295X2 on a single card. It's interesting. Because it will bring down all 28 nm cards.

BTW...Intel HD is doomed again..if a mid-high class chip can do that, wondering how a low tier card can do.

If they don't rebrand any desktop cards and are all in the new uArch... Intel HD would be outclassed hard.
amd never did good on firestrike i dont really think its the best example to compare it lol
 
amd never did good on firestrike i dont really think its the best example to compare it lol

No SINGLE benchmark is good for any proper comparison. In other words, we need to wait.

Totally OT, that reminds me of a Mitch Hedberg joke: I saw this wino, he was eating grapes. I was like "Dude, you have to wait."
 
Okay, I thought I would wait until the Big Pascal and Big Vega drop next year to replace my 2 R9 290s in rig 2 below.

I don't mine with the 290s, just game and both rigs are custom water cooled so I have to buy a new gpu block no matter what card I go with.

RS and I have been discussing options and an option came up about selling the 2 R9 290s, moving the GTX980 TI to replace them and replace the 980 TI with a 1080 in the 5960x rig (no need to "upgrade" to 6950x).

Your thoughts?

Just sell everything and get two 1080s. I'm seriously debating this route myself. Instead of passing my 980 Ti to the GF, just sell it, and get two 1080s one for each of us.

If the DX12 revolution is as severe for the next two years, I'll give my nephews the water cooled 290X and keep the 780 Lightning in the basement PC (I basically only play DX11 or lower games down there anyways).
 
Just sell everything and get two 1080s. I'm seriously debating this route myself. Instead of passing my 980 Ti to the GF, just sell it, and get two 1080s one for each of us.

If the DX12 revolution is as severe for the next two years, I'll give my nephews the water cooled 290X and keep the 780 Lightning in the basement PC (I basically only play DX11 or lower games down there anyways).[/

nvm
 
Okay, I thought I would wait until the Big Pascal and Big Vega drop next year to replace my 2 R9 290s in rig 2 below.

I don't mine with the 290s, just game and both rigs are custom water cooled so I have to buy a new gpu block no matter what card I go with.

RS and I have been discussing options and an option came up about selling the 2 R9 290s, moving the GTX980 TI to replace them and replace the 980 TI with a 1080 in the 5960x rig (no need to "upgrade" to 6950x).

Your thoughts?

I agree that in your case, choosing a new GPU over a 6950x is a much more exciting and fun proposition. Unless you can use 10 cores on a regular basis, I don't think a 6950X will make you smile very much, but a new 1080 just might do the trick!
If you get the new card, you'd be in a great position to do direct comparisons between 980ti and 1080 performance. That right there might be VERY fun for you to do and I bet you would enjoy that, OCing both of them and comparing the two side by side etc. Both CPU's are haswell based so that would be really great for comparing the two cards with. Sounds like fun. I say do it if you have the desire.
 
Waiting for Vega here.

Will not buy another nVidia card this cycle (currently running 2 x 670) for two reasons. One of which is the "performance falling off when a new generation comes out" syndrome. The other is the lack of FreeSync.
 
I get 9607 graphics score with a single 1440core~ 8000memory 980ti and 17100 graphics score with SLI on a tired old horse i7 980x:

http://imgur.com/a/rr6h2

Suddenly that 10k score from the 1080 doesn't seem much of an improvement..

When you think again and realize how much higher it will likely be with driver improvements, and that the 1080 probably isn't the top of the line, it seems pretty good. 🙂

We likely have a 1080ti and then the top dog card, both above the 1080.
 
I get 9607 graphics score with a single 1440core~ 8000memory 980ti and 17100 graphics score with SLI on a tired old horse i7 980x:

http://imgur.com/a/rr6h2

Suddenly that 10k score from the 1080 doesn't seem much of an improvement..

That echoes other comparisons like below. Given that, it's not unusual or terrible in context. Your outgoing 980ti at max or near max clock just about equals the new card at stock. That's what's expected no?

https://www.reddit.com/r/nvidia/comments/4i0avr/benchmark_1080_vs_980ti_watercooled_on_3dmark11/
 
Just sell everything and get two 1080s. I'm seriously debating this route myself. Instead of passing my 980 Ti to the GF, just sell it, and get two 1080s one for each of us.

If the DX12 revolution is as severe for the next two years, I'll give my nephews the water cooled 290X and keep the 780 Lightning in the basement PC (I basically only play DX11 or lower games down there anyways).

I suggested that route too. He has a special situation though since reference 1080 is now a $699 card and most likely early water blocks will be made for it. There isn't enough info yet about $599 after-market cards. Since the 980Ti already has a waterblock, it makes upgrading from 980Ti OC to 1080 less of a slam dunk for him. Getting rid of 290s is another story since they rely on CF to perform well which would make 1080 OC s hands down superior option overall.

Reselling 290s and upgrading to 1080, still leaves room down the line to sell the 980Ti and get a Big Pascal.

I also fear NV might do a repeat of Kepler generation:

680 = 1080
Titan = cut down Titan Pascal (3584 CCs)
780 = cut down GP100 (similar to 780Ghz)
Refreshed 1070/1080 (we could see 1070Ti and 1085 adopting faster GDDR5X and GPU clocks)
780Ti = full GP100

Given how much ahead NV will be with 1080 and Vega 10 so far away, there is no reason for NV to release a full chip GP100 in Q1 2017 either. NV might drag this generation way out since 70% of their install base is using pre-Maxwell GPUs leaving them A LOT of room to capture upgraders with the 1st, 2nd and 3rd wave of Pascal chips.

On other forums, members are voicing similar predictions.
 
Last edited:
Just sell everything and get two 1080s. I'm seriously debating this route myself. Instead of passing my 980 Ti to the GF, just sell it, and get two 1080s one for each of us.

If the DX12 revolution is as severe for the next two years, I'll give my nephews the water cooled 290X and keep the 780 Lightning in the basement PC (I basically only play DX11 or lower games down there anyways).

Wrong move.
Sell the 980Ti, get Two 1070s
So now you're in the same position, but you don't suffer Maxwell detereotation.
Then, when 1080ti comes out, you get a 1080ti for yourself.

You'll pay a little extra for the two 1070s, but at least you'll have a 1080Ti for yourself, and the GF has SLI 1070s. If your GF doesn't like it, I can be your new GF.
 
I suggested that route too. He has a special situation though since reference 1080 is now a $699 card and most likely early water blocks will be made for it. There isn't enough info yet about $599 after-market cards. Since the 980Ti already has a waterblock, it makes upgrading from 980Ti OC to 1080 less of a slam dunk for him. Getting rid of 290s is another story since they rely on CF to perform well which would make 1080 OC s hands down superior option overall.

Reselling 290s and upgrading to 1080, still leaves room down the line to sell the 980Ti and get a Big Pascal.

I also fear NV might do a repeat of Kepler generation:

680 = 1080
Titan = cut down Titan Pascal (3584 CCs)
780 = cut down GP100 (similar to 780Ghz)
Refreshed 1070/1080 (we could see 1070Ti and 1085 adopting faster GDDR5X and GPU clocks)
780Ti = full GP100

Given how much ahead NV will be with 1080 and Vega 10 so far away, there is no reason for NV to release a full chip GP100 in Q1 2017 either. NV might drag this generation way out since 70% of their install base is using pre-Maxwell GPUs leaving them A LOT of room to capture upgraders with the 1st, 2nd and 3rd wave of Pascal chips.

On other forums, members are voicing similar predictions.
can you post the link again? i had yesterday but i lost it
 
That is a low perf/w jump, considering AMD is claiming up to a 2.5 perf/w performance increase. Nvidia making their architecture more GCN-like has increased power usage.

Maxwell was already considerably more power efficient that most AMD cards, so a bigger increase in perf/w should come for AMD.
 
I get 9607 graphics score with a single 1440core~ 8000memory 980ti and 17100 graphics score with SLI on a tired old horse i7 980x:

http://imgur.com/a/rr6h2

Suddenly that 10k score from the 1080 doesn't seem much of an improvement..

You have to remember that the 1080 will be able to overclock as well, and it seems from the presentation that it's just as good at doing so as the 980 Ti, so once you factor that in the 1080 will still have it's ~25% performance gains.

Another thing that has to be looked at is the architectural improvements that are going to allow the 1080 to have better DX12 performance. If DX12 adoption rate improves, the 980 Ti has a more limited useful lifespan.
 
Wrong move.
Sell the 980Ti, get Two 1070s
So now you're in the same position, but you don't suffer Maxwell detereotation.
Then, when 1080ti comes out, you get a 1080ti for yourself.

You'll pay a little extra for the two 1070s, but at least you'll have a 1080Ti for yourself, and the GF has SLI 1070s. If your GF doesn't like it, I can be your new GF.

Replacing 980ti's for two 1070's, as soon as I saw you say it, struck me as a brilliant idea to stay around the same performance without spending hardly anything at all (or actually pocket some money) to do it and avoid driver issues. Then get 1080ti later.
But, even with a technically free move to a safer GPU solution, I have to ask myself, "Even if free, is it worth the hassle? What games are lacking in performance?" The answer is none.
By the time I gain any performance benefit from having 1070's, I will be wanting either a single or dual big die cards for BF5. I think at this point someone would have to PAY ME to replace my GPU's. Rocket league is running just fine LOL.
 
When you think again and realize how much higher it will likely be with driver improvements, and that the 1080 probably isn't the top of the line, it seems pretty good. 🙂

We likely have a 1080ti and then the top dog card, both above the 1080.

Depends if we want to lower our standards for performance and price/performance.

670 beat 580 (Titan X) by ~ 20% at 1600P, while 680 beat it by 30% and 7970Ghz beat it by 41%!

perfrel_2560.gif


Now NV is (1) raising prices on blower cards (2) keeps raising prices on mid-range cards $499 680 -> $549 980 -> $599-699 1080. Also, 1070 is getting the worst neuter of any x70/x75 series card I can remember. It may be impossible to overclock 1070 to 1080 speeds and it's doubtful that 1070 won't be 20% faster than the Titan X unlike 670 vs. 580. Even though the 1070 sounds amazing on paper by beating Titan X, most people bought after-market 980Ti not the Titan X. As far as comparing 1070 directly to 1080, the 1070 looks more like a 1060Ti if it has 1920-2048 CCs and regular GDDR5. Big GIMP for an x70 series card.

That means objectively, NV is delivering less of a performance leap that Kepler did over Fermi, and raising prices at the same time. I don't see how that constitutes "pretty good". What I am seeing instead is a 3rd consecutive generation of milking the mid-range, bifurcating a generation and essentially forcing gamers who are in-line to upgrade to either pay MORE for next gen mid-range performance OR wait another 9-12 months for the real flagship. That means AMD/NV set up new generations now in a way that a gamer can't really win. It seems the best way to upgrade now is to dump old cards ASAP/time resale and pass on the outdated and overpriced old hot potato. The $199 and below segment is also very weak nowadays. AMD and NV are trying desperately to get gamers to make the new mid-range sweet-spot the $300-400 price range, not $200-250 as was the case in the past.

can you post the link again? i had yesterday but i lost it

Ya, because that slide I linked is now forbidden from being viewed for some reason or it was taken down.

Here are some slide decks - you should find it under Growth Opportunity slide:
http://xtreview.com/computer-hardware93.htm

I think at this point someone would have to PAY ME to replace my GPU's. Rocket league is running just fine LOL.

In that case, you could do a hybrid strategy to minimize your losses. Sell 1 980Ti for $480-$500 next week, keep the remaining 980Ti. Use the 980Ti for all the easy games that don't give it trouble. Then since your big game upgrades center around BF, you can either sell the 2nd 980Ti around October 21, 2016, and pick up 2 after-market 1080s such as EVGA Classified, G1s or MSI Lightnings that may hit 2.25-2.3Ghz air overclocks OR you surely will have plenty of $380 after-market 1070s available by then. Since you 'sold' 980Ti now for $480-500, even if you sell the 2nd 980Ti for $380 around October 2016, if you buy a 2nd 1070 for $380, you have $100 in your pocket ($480/500 less $380 = $100-120). The first strategy gives you time to upgrade to much faster after-market 1080 AIB cards, while the 2nd strategy also has little risk since by then you'll see how much better, if at all, 1070 is over 980Ti in DX12 games. In both cases, you simply lock in the $480-500 from selling the 2nd 980Ti right now. Another potential benefit is you may realize a single fast card is enough for you and won't need to spend $1300-1400 on Big Pascal cards in 2017.

Is it worth the hassle for you? Maybe not but there are options. 😎
 
Last edited:
I personally wouldn't touch the 1070s until I had the full rundown on them. And this could take a while. I would want to know how much is truly chopped out of the chip i.e Vram, ROPS, etc

The flip-flopping for the newer tech does indeed sound great on paper but it's not always that cut and dry.

1. You need to find a respectable buyer or else the card might come bouncing back
2. Waiting around to finally purchase that new tech. This can be aggravating sometimes.
3. And when you get your new tech is it performing properly or is there some kind of malfunction. Coil whine, Driver issues, etc.
 
I don't think you'll be able to get anywhere near $500 for a 980 Ti now. Just checked completed listings and prices on eBay have dropped to $400-450, even the EVGA Classified model was selling in that range. The only >$500 ones were brand new.

Still a good time to sell though, I think prices will settle down to $350 or so once the 1070 is released.
 
Wrong move.
Sell the 980Ti, get Two 1070s
So now you're in the same position, but you don't suffer Maxwell detereotation.
Then, when 1080ti comes out, you get a 1080ti for yourself.

You'll pay a little extra for the two 1070s, but at least you'll have a 1080Ti for yourself, and the GF has SLI 1070s. If your GF doesn't like it, I can be your new GF.

Haha :thumbsup:

Truth be told, call me a sheep, whatever, I love buying new computer parts. PC Building and Gaming are my bigger hobbies, where as other people drop $600-700 on a good set of Golf Clubs they use seasonally, I got no qualms with that on a PC part I'll use daily.

So depending on the gap, I'll probably get her a 1070, sell it once 1080 Ti is out and toss her my 1080 and upgrade to 1080 Ti.

Keep in mind, this is exactly how I did it with AMD/ATI. I'd upgrade every refresh/generation. They lost a damn good customer, but I doubt they care. I'm just an ant to them anyways.
 
Wrong move.
Sell the 980Ti, get Two 1070s
So now you're in the same position, but you don't suffer Maxwell detereotation.
Then, when 1080ti comes out, you get a 1080ti for yourself.

You'll pay a little extra for the two 1070s, but at least you'll have a 1080Ti for yourself, and the GF has SLI 1070s. If your GF doesn't like it, I can be your new GF.
That's not a bad idea, I doubt if that's the option he'll chose, but you never know :sneaky:

I think the 3 Pascal flagships, as RS mentioned, is a real possibility but it also depends on how far out Volta is & the die size of Vega or Navi. I included Navi because if that, multi GPU via interposer, is the route AMD's taking then Nvidia may also follow suit & we could see the real big die strategy coming to an end.
 
Back
Top