NVIDIA GeForce GTX 1080 Founders Edition Review @ [H]

You say 30-35% as if its never lower. Some reviews have it averaging lower than that and in some games its only 15-20% better. The 980 ti also has more overclocking headroom so at the end of the day it would be a pathetically small upgrade when looking at oced 980 ti to oced 1080. I still dont get all this praise as in my opinion this card is a joke for a next gen product at 700 bucks.

The price is a bit high, but for a 980 replacement is does bring pretty good performance. You have to remember that this isn't the 980ti or Titan X successor.
 
The price is a bit high, but for a 980 replacement is does bring pretty good performance. You have to remember that this isn't the 980ti or Titan X successor.
Yes I just referring to this overpriced founders edition for now. Still I am worried that any decent cards such a the MSI Gaming version may end up priced closer to 700 than the "msrp" of 600.
 
Yes I just referring to this overpriced founders edition for now. Still I am worried that any decent cards such a the MSI Gaming version may end up priced closer to 700 than the "msrp" of 600.

You know that's going to happen, especially with reference setting the bar at $700. It wouldn't make sense to create for example an ACX 3.0 card with extra phases for just $600 as they will at least ask for $630-$650 for that and then the kick ass versions(e.g. FTW/Kingpin) with 2 x 8 pin + phases and custom cooling will easily cross $750.

NVIDIA should have priced FE at $600 (or IMO $499) and that would've pressured everyone else to start there. Now that pressure has been pretty much lifted.
 
You know that's going to happen, especially with reference setting the bar at $700. It wouldn't make sense to create for example an ACX 3.0 card with extra phases for just $600 as they will at least ask for $630-$650 for that and then the kick ass versions(e.g. FTW/Kingpin) with 2 x 8 pin + phases and custom cooling will easily cross $750.

NVIDIA should have priced FE at $600 (or IMO $499) and that would've pressured everyone else to start there. Now that pressure has been pretty much lifted.



Tom Petersen from Nvidia discussed the Founder's Edition a bit in this video (around 2:09:00), he says they intended it to be in the middle of the product stack price/quality wise.

So yeah, I think the FE is not a price ceiling of any sort, we will see designs that go above it and as a result cost more.
 
You know that's going to happen, especially with reference setting the bar at $700. It wouldn't make sense to create for example an ACX 3.0 card with extra phases for just $600 as they will at least ask for $630-$650 for that and then the kick ass versions(e.g. FTW/Kingpin) with 2 x 8 pin + phases and custom cooling will easily cross $750.

NVIDIA should have priced FE at $600 (or IMO $499) and that would've pressured everyone else to start there. Now that pressure has been pretty much lifted.
Yeah if you watch the pcper video, Tom says that some cards will be priced higher than the founders day card and that the founders day card is essentially priced in the middle of the stack. If that turns out to be true then we are pretty screwed.

EDIT: Lol I was too slow and x3sphere already posted the link for the vid...
 
The temperatures on this card are abysmal. I had hoped that being a Founder's Edition cooler plus smaller process the temperatures would be much lower. Here is hoping custom cooled cards are hitting 65 C or something.
 
The temperatures on this card are abysmal. I had hoped that being a Founder's Edition cooler plus smaller process the temperatures would be much lower. Here is hoping custom cooled cards are hitting 65 C or something.
And since they already put the vapor chamber cooler on the 1080 then I wonder what the heck the big boy 1080ti will use...
 
Yeah if you watch the pcper video, Tom says that some cards will be priced higher than the founders day card and that the founders day card is essentially priced in the middle of the stack. If that turns out to be true then we are pretty screwed.

EDIT: Lol I was too slow and x3sphere already posted the link for the vid...



Tom Petersen from Nvidia discussed the Founder's Edition a bit in this video (around 2:09:00), he says they intended it to be in the middle of the product stack price/quality wise.

So yeah, I think the FE is not a price ceiling of any sort, we will see designs that go above it and as a result cost more.



Haha both of you posted the same message but yeah that interview spells it out really clearly. This was supposed to be a great way for NVIDIA to move the price yardstick up a few notches without anyone really noticing but it backfired on them because it was handled wrong. What I got from the video is this:

  • Shit Tier Editions with potentially lesser quality cooling and PCB: $599 aka Galax 1080
  • Middle Tier aka Founders Edition: $700
  • Custom cooling + maybe extra phases aka I'm Not a Peasant Edition: $730-$750
  • Kick ass editions aka The Whole Enchilada Edition with 2 x 8 pin, custom cooling, extra phases etc: $780-$800
That's my prediction.


And since they already put the vapor chamber cooler on the 1080 then I wonder what the heck the big boy 1080ti will use...

Hopefully NVIDIA straps an AIO on it like AMD did for Fury X.
 
Last edited:
I'm a big fan of Eurogamer/Digital Foundry video reviews so I figured I'd toss these up for those that also enjoy them:





Pay attention to the Hitman and AotS DX 12 benchmarks. Rich tries to be diplomatic about Pascal's shit performance there but the numbers speak for themselves. If AMD can move Vega up to October and deliver, then damn, I bet Pascal is in trouble.
 
Last edited:
Digital Foundry makes great videos though, pretty much the best as their videos give a lot of info, much more than just basic graphs.

Only Ashes has Async On/Off toggle (in the .ini file!), so its the only one where you can test NV's claims. Computebase.de and some other sites did that and they found Pascal 1080 doesn't suffer regression when Async is On, as Maxwell does.

So NV's claims of Async Compute is not true, but Pascal seem to cope better graphics & compute switching or preemption.

The reference card at $699 is a sick joke though. What it means and we'll see it, are AIB custom models will not be $599, but start around $699 onwards. Who the heck is gonna price their 8/10 phase custom board with better components and cooling cheaper? It will come out of the box with 2Ghz boost clocks, and can crank to much higher than the mediocre Fury X-like overclockability of the reference 1080.
 
Oh and pcgameshardware did a nice video where they replaced the reference cooler with a custom cooler, and they still could not maintain a high 2.1ghz OC. After a few minutes it throttles back down, hitting the TDP limits despite running cool.

I think hardware.fr also did some OC clock monitoring and they also found it drop back down after a few minutes. That's really crap for a $100 premium over MSRP.
 
Oh and pcgameshardware did a nice video where they replaced the reference cooler with a custom cooler, and they still could not maintain a high 2.1ghz OC. After a few minutes it throttles back down, hitting the TDP limits despite running cool.

I think hardware.fr also did some OC clock monitoring and they also found it drop back down after a few minutes. That's really crap for a $100 premium over MSRP.

Temp doesn't help much with power limits. Undervolting can... No idea if that is feasible.

Everyone seems to ignore the card ramp up to temp THEN ramp the fan higher. Pretty sure it could do 67C if you set it to that, fan might be pretty loud though.
 
I'm a big fan of Eurogamer/Digital Foundry video reviews so I figured I'd toss these up for those that also enjoy them:





Pay attention to the Hitman and AotS DX 12 benchmarks. Rich tries to be diplomatic about Pascal's shit performance there but the numbers speak for themselves. If AMD can move Vega up to October and deliver, then damn, I bet Pascal is in trouble.

NVIDIA actually has substantially better performance in Episode 2 of Hitman than AMD now in DX11, and still just ahead of Fury X with a 980ti in DX12.
Hitman (2016) Episode 2: DX12 mit bis zu 60 Prozent Leistungsplus - und Problemen
Caveat being 970 suffers though (possibly the memory *shrug*)

Anyway something is not right with episode 1, which even pcgameshardware showed being crud on all NVIDIA cards.
Strange though in episode 2 they identified that AMD cards have a very slight visual quality issue in comparison to NVIDIA with DX12+"memory protection"; they provided a picture showing the difference and it can be seen, but not sure if AMD has resolved this with a driver or a game patch yet or ever will.
Cheers
 
Last edited:
Who cares about firestrike? Go read computerbase's review, they did OC vs OC in games.
And problem is that the OC behaviour-variables are NOT identical between Maxwell and Pascal, Boost3 is much more dynamic and fine grained requiring a more thorough approach to OC or better use of OC utilities such as the beta EVGA one.
TPU managed a 10% OC (which translated into 10% boost in BF3) by doing it the old fashioned way with the 1080, meaning they are not playing with fan speed/power target/voltage-to-frequency/etc.
Cheers
 
I am impressed!

1080 is ~75% faster than 980 and ~35% faster than 980Ti at 4k. Overclock at 2100mhz was almost twice as fast as 980Ti SLI. That's a very nice boost! Still not buying it ;)
BUT, I am looking forward to 1080Ti which should be blazing fast. If it has HBM2 then 4k performance should be even better. Hoping here, and then I'm getting two of those :D
 
Thanks. I wonder if this only for DX11+ fullscreen applications or enabled 10 bit output for content creation. Nvidia used to allow this on Quadro cards only.
I'd be nearly willing to bet it's only for DX again as it has been over the last 3 generations. No OGL no nothing else, so no doom4 10 bit glory. Or else it'd work in CAD programs..


This launch really made me appreciate what AMD did with the 7970. No power limits, no gimped bullshit, just decent overclocking chips with thermal limits. Only needed volt modding if ln2/extreme chilled water etc...

Edit: to all the 'twice as fast as 980ti' people, I don't know anyone who has not also OC'd the 980tis.... which will definitely even it out. Let alone these mythical 2.1ghz cards are power limited..

We need double-blind GPU reviews in order to get objectivity.

Some HiFi reviewers have resorted to this. Could be a great approach to investigate for [H].. no doubt they already have though ;)
 
Last edited:
As someone still running a 770 GTX (2GB edition; major mistake there in hindsight), I'm almost certainly getting one of these, despite the price premium. Was hoping for HBM2, but the performance of GDDR5x is clearly there.
 
Have there been any reviews comparing the 1080 to an overclocked 980ti?

As it stands, I don't see any compelling reason to upgrade from my 980ti until the 1080ti becomes available, hopefully sooner rather than later.
 
Very impressive. I will wait to see how Polaris 10 performs. If It is indeed around 390x speeds than, I will just grab an aftermarket 1080. Maybe an MSI lightening or EVGA Classified.
 
Correct question.
Check here for numbers of the 1080GTX 8GB Vs TitanX 12GB or R9 295x2 8GB : NVIDIA GeForce GTX 1080 8 GB Review

This shows that the 2x[4gb] 295x2 did rather well at 4k. How much a GPU uses and what it absolutely needs before dropping frames is hard to determine, especially with dynamic memory usage:
VRAM Usage - Tom Clancy's The Division Gameplay Performance Review

Really, bandwidth might be more important than capacity for keeping the frames consistent. The FuryX trails the 980ti at 1440p but usually beats it at 4k. In the above TPU article, the 295x2 pulls ahead of the Titan X at higher resolutions. Clearly 4 GB is STILL enough, so long as you have good memory allocation and bandwidth. 8 GB is nice to have, but I would rather of seen 384 bit GDDR5x with 6 GB personally. GDRR5x is immature right now, as we see that the 980ti beats it in bandwidth.

Folding@home mentioned above. Can you please give the PPD for this card?
 
Pay attention to the Hitman and AotS DX 12 benchmarks. Rich tries to be diplomatic about Pascal's shit performance there but the numbers speak for themselves. If AMD can move Vega up to October and deliver, then damn, I bet Pascal is in trouble.

In trouble how? Whatever AMD comes out with next, Nvidia will counter with the Titan and we are right back where AMD started. Behind.
This is a never ending loop every year. Like Kyle always says, you can only go by what's available now. Right now, the 1080 offers the best gameplay experience.


The price is a bit high, but for a 980 replacement is does bring pretty good performance. You have to remember that this isn't the 980ti or Titan X successor.

Nah, people will continue to compare this to the Ti and Fury because they lack understanding. Big chip vs small chip isn't a fair comparison.
The 1080 nearly doubles the performance of the 980. Comparing similar chip size, this is an incredible performance leap to where it does beat the older larger chip cards.

If your a big chip customer, then wait for the big Pascal chip to compare. But it's silly to shame a smaller chip card that is clearly out performing the big boys from last gen.
 
That this card is a replacement for the 980 non-ti it performs really well. I'm waiting for whatever the 1080ti model or the Titan XX model (hopefully it's not 2 grand)
 
In trouble how? Whatever AMD comes out with next, Nvidia will counter with the Titan and we are right back where AMD started. Behind.
This is a never ending loop every year. Like Kyle always says, you can only go by what's available now. Right now, the 1080 offers the best gameplay experience.




Nah, people will continue to compare this to the Ti and Fury because they lack understanding. Big chip vs small chip isn't a fair comparison.
The 1080 nearly doubles the performance of the 980. Comparing similar chip size, this is an incredible performance leap to where it does beat the older larger chip cards.

If your a big chip customer, then wait for the big Pascal chip to compare. But it's silly to shame a smaller chip card that is clearly out performing the big boys from last gen.
Maybe you lack understanding that people are comparing purchase options. It makes all the sense in the world to attempt to compare the top end models that are closest to each other in performance. That helps asses value and savings against each model. Comparing to every single legacy model is a waste of time. Any GPU upgrade would be an improvement for an older model.

Comparing vs chip size is purely an academic exercise that has no bearing on actual purchase decisions.
 
Maybe you lack understanding that people are comparing purchase options. It makes all the sense in the world to attempt to compare the top end models that are closest to each other in performance. That helps asses value and savings against each model. Comparing to every single legacy model is a waste of time. Any GPU upgrade would be an improvement for an older model.

Comparing vs chip size is purely an academic exercise that has no bearing on actual purchase decisions.

Purchase options? lol

Who the hell is deciding whether or not to buy a new Ti/Titan/Fury over a GTX 1080 right now?
If you're talking about used cards, then price will bottom out on the Ti's and Fury's soon, but still that's not a fair comparison because it doesn't represent the actual retail value.
 
What's a good expectation of when we start to hear specifics of big(ger) Pascal? I think it may depend on what AMD releases. 2-4 months?
 
I kinda feel sorry for people with a 7 series card waiting to buy this new edition. I would personally feel like waiting even longer to see what AMD does and whether nvidia just jacks prices up sky high before then. I know hardly any DX12 stuff is out but when you buy one of these new cards that should be a significant consideration. As a 980 owner one thing I can say is gsync is well worth having for times like these.
 
What's a good expectation of when we start to hear specifics of big(ger) Pascal? I think it may depend on what AMD releases. 2-4 months?

I'll tell you some insider info I have, that I assure you is 100% accurate.

You will certainly hear about the Pascal Titan/Ti within 9 months. In fact, one could argue, you're hearing about right now.
 
Quick question to folks who are better versed in these things. Right now I have 3 1080p 3d monitors in nSurround, but there's no display port connector. These updated display ports do you think would be able to connect the other two with 'non active' dongles? What would you project 5760 x 1080 performance to be? ( I know 4k is all the rage now but I have 3 perfectly good monitors and I kinda like it for flying and driving games.... like looking out the side windows)
 
What's a good expectation of when we start to hear specifics of big(ger) Pascal? I think it may depend on what AMD releases. 2-4 months?

We already know what is in "Big Pascal". It is 610mm^2, 3584 CUDA cores, 15.3 billion transistors. It makes GP104 / GTX 1080 look like a peon. Inside Pascal: NVIDIA’s Newest Computing Platform

The question to ask is if there will even be a GTX 1080Ti (there wasn't a GTX 680Ti, which is the last time we had a process node shrink), and if there will be, will it will use GP100?
If the rumors are true (and many for Pascal have been), then there could be a chip inbetween the GP100 and GP104, the GP102: NVIDIA Pascal GPU codenames revealed
 
We already know what is in "Big Pascal". It is 610mm^2, 3584 CUDA cores, 15.3 billion transistors. It makes GP104 / GTX 1080 look like a peon. Inside Pascal: NVIDIA’s Newest Computing Platform

The question to ask is if there will even be a GTX 1080Ti (there wasn't a GTX 680Ti, which is the last time we had a process node shrink), and if there will be, will it will use GP100?
If the rumors are true (and many for Pascal have been), then there could be a chip inbetween the GP100 and GP104, the GP102: NVIDIA Pascal GPU codenames revealed
I believe nVidia will release a big Pascal, Ti, or whatever it will be called, once AMD release a new flagship GPU by the end of this year or next year. They'll need something to compete with AMD rather than not releasing anything until the next architecture.
 
We already know what is in "Big Pascal". It is 610mm^2, 3584 CUDA cores, 15.3 billion transistors. It makes GP104 / GTX 1080 look like a peon. Inside Pascal: NVIDIA’s Newest Computing Platform

The question to ask is if there will even be a GTX 1080Ti (there wasn't a GTX 680Ti, which is the last time we had a process node shrink), and if there will be, will it will use GP100?
If the rumors are true (and many for Pascal have been), then there could be a chip inbetween the GP100 and GP104, the GP102: NVIDIA Pascal GPU codenames revealed

Yes that would be pretty crazy; for some reason i doubted it would come to the consumer market and thought they were going to keep separate chips (aKa GP102 = big pascal). In any case, my speculation won't help; market pressure to show more of their cards (no pun intended) is the only way we'll see its full potential. Here's to AMD's next generation of GPUs!
 
Quick question to folks who are better versed in these things. Right now I have 3 1080p 3d monitors in nSurround, but there's no display port connector. These updated display ports do you think would be able to connect the other two with 'non active' dongles? What would you project 5760 x 1080 performance to be? ( I know 4k is all the rage now but I have 3 perfectly good monitors and I kinda like it for flying and driving games.... like looking out the side windows)


Personally I would not use passive dongles......
 
What's a good expectation of when we start to hear specifics of big(ger) Pascal? I think it may depend on what AMD releases. 2-4 months?

I wouldn't be overly surprised if we hear about it by the end of the year. Though, it will of course depend on what AMD does. If AMD manages to surpass everyone's predictions and release an amazing card at a really cheap price it will make Nvidia want to react and we might see a nice price drop on these cards with a new performance leader following closely after. However, if predictions on AMD's upcoming card are right it might take a while for something better to appear, especially at a more reasonable price.
 
Quick question to folks who are better versed in these things. Right now I have 3 1080p 3d monitors in nSurround, but there's no display port connector. These updated display ports do you think would be able to connect the other two with 'non active' dongles? What would you project 5760 x 1080 performance to be? ( I know 4k is all the rage now but I have 3 perfectly good monitors and I kinda like it for flying and driving games.... like looking out the side windows)

You've read my mind. I have exactly the same situation with 3 3D monitors that I won't to keep. Right now with 3 GTX 680s I simply plugged them up to dual link DVI ports. But that won't be possible using the 1080/1070, not officially with only 2x SLI support. It should be possible using an active DP to dual link DVI adapter like this one: Amazon product ASIN B00DYRQXMK. We'd need three as all of the monitors in 3D Surround have be connected the same way, you can't mix and match DP and DVI. Results on these look a bit mixed. Plan on ordering three though and hoping these work. If anyone else has experience with DP to dual link DVI your input would be greatly appreciated.
 
Purchase options? lol

Who the hell is deciding whether or not to buy a new Ti/Titan/Fury over a GTX 1080 right now?
If you're talking about used cards, then price will bottom out on the Ti's and Fury's soon, but still that's not a fair comparison because it doesn't represent the actual retail value.
You seem caught up on this extraneous "fairness" argument.

Many people will be waiting to upgrade until the new series of cards comes out. They will either decide to outright buy the new card, if it is substantially better enough, or get a discount on the previous best cards. Either way, the [H] comparisons have a great deal of value. It is not worth wasting time comparing cards, that are already benchmarked on these games, that will not be in the purchase discussion.

This is not an arithmetic pissing contest. Everything doesn't have to be the same chip size, clock speed, or price in order to be valuable for comparison purposes.
 
Back
Top