NVIDIA GeForce GTX 1080 Founders Edition Review @ [H]

thelead said:
I'm already questioning how they hit 2100Mhz at 67C in that doom demo...
Click to expand...

The game was probably only pushing the GPU to 60% - 70% usage I'm guessing.


Dapperdan said:
So for those who have 980 Tis, is this worth considering an upgrade?
Click to expand...

You'll hear a lot of opposing views. It just comes down to how much you value 20% - 30% extra performance over an overclocked 980 Ti.
 
Thanks for the review. From the review:

If this game is a better example of Async Compute and its benefits under DX12 it doesn't seem to be a large improvement in the grand scale of things. The improvement advantage doesn't offset the large performance improvement GeForce GTX 1080 Founders Edition natively brings. The best gaming experience in DX11 or DX12 in this game is the new GeForce GTX 1080 Founders Edition.
Click to expand...

I think the conclusion is odd. The benefit of async compute is discounted because it doesn't enable a last gen 28nm card to beat the latest 16nm card? This seems to be an impossibly high bar to set. The question is not whether Fury X with async compute could match Pascal - I don't think anyone expected it would - it's whether Polaris/Vega will show similar gains with async compute that its predecessor does. And whether Pascal's reduced negative gain versus Maxwell is the result of pure brute force or architectural changes that might eventually lead to Pascal showing gains with DX12/async.

Brent_Justice said:
I think the poor DX12 performance is more to do with the application. In terms of Tomb Raider, its slower on AMD and NVIDIA in DX12. Hitman freezes on me. In AoTS I think the game has been so heavily optimized for AMD, and not NVIDIA's architecture. That is my theory. We need more real DX12 games built from the ground up to determine DX12 performance.
Click to expand...

Does the theory of AMD-optimization for AoTS have any supporting evidence? The developers have spoken extensively about this topic and they have denied there is any AMD-unique optimization in their game:

Personally, I think one could just as easily make the claim that we were biased toward Nvidia as the only ‘vendor’ specific code is for Nvidia where we had to shutdownasync compute. By vendor specific, I mean a case where we look at the Vendor ID and make changes to our rendering path. Curiously, their driver reported this feature was functional but attempting to use it was an unmitigated disaster in terms of performance and conformance so we shut it down on their hardware. As far as I know, Maxwell doesn’t really have Async Compute so I don’t know why their driver was trying to expose that. The only other thing that is different between them is that Nvidia does fall into Tier 2 class binding hardware instead of Tier 3 like AMD which requires a little bit more CPU overhead in D3D12, but I don’t think it ended up being very significant. This isn’t a vendor specific path, as it’s responding to capabilities the driver reports.
Click to expand...

[Various] Ashes of the Singularity DX12 Benchmarks - Page 121

It's certainly true we need more working DX12 games for benchmarking but while we wait for that and unless there's evidence that challenges what the developers have stated, don't see what justifies attributing the results in Ashes of the Singularity to AMD optimization.
 
thelead said:
I'm already questioning how they hit 2100Mhz at 67C in that doom demo...
Click to expand...
PCper did test overclocking and increased the power limit, hitting 2025Mhz at 85C. I don't know that I'd be comfortable letting my $700 videocard sit at that temperature, personally. AIB cooling is a must.
 
Some other reviewers had some overclocked 1080's running at 2.1/5.6 on the founders cooler. They did note that the default fan curve on the 1080 was woefully tuned down to be a quieter card.

When OC'd to that level, they had performance on par with 980ti SLI.
 
schizo said:
30% gain versus a reference 980ti. AIB 980tis overclock like mad, and 1500Mhz is not uncommon.

Of course the 1080 will overclock too, but from the [H] review, looks like we need AIB cooling for that-- his reference 1080 hit 80C at normal boost clocks.
Click to expand...

That is 21%-30% across the board with beta drivers....
It will only increase with drivers done like they are for the 980/980ti.

Still not worth upgrading if still own a 980ti and happy with current gaming and can wait, but that is a good performance jump for the replacement to the 980.
Cheers
 
I have yet to see anyone review this card for VR. VR is the only reason I would consider buying one of these now and it would be nice to see it vs a 980ti for VR. Let me know if anyone comes across a thorough review that included VR comparisons.
 
Vashypooh said:
Some other reviewers had some overclocked 1080's running at 2.1/5.6 on the founders cooler. They did note that the default fan curve on the 1080 was woefully tuned down to be a quieter card.

When OC'd to that level, they had performance on par with 980ti SLI.
Click to expand...
I'm slowly grinding my way through all the reviews-- links?

CSI_PC: Totally agree, if you already have a 980ti this is an underwhelming upgrade for sure.
 
schizo said:
30% gain versus a reference 980ti. AIB 980tis overclock like mad, and 1500Mhz is not uncommon.

Of course the 1080 will overclock too, but from the [H] review, looks like we need AIB cooling for that-- his reference 1080 hit 80C at normal boost clocks.
Click to expand...

my volt modded 980ti can run 1500mhz, and still loses to a stock clocked 1080 in 3dmark, I just barely broke 10000 graphics score in 3dmark extreme once, and mind you i couldnt run that 1550 core clock for more than a few minutes, without crashing. i run 1480 in games, and its pretty solid, but can can definately get noisy. I have a copper 2 fan aftermarket cooler too btw. My temps damn near 80C at 90% fan speed.

so i get more performance with less head ache, and can still overclock for another 10-15% performance.

sold!
 
GTX 1080 OC results

screenshot_www_computerbase_de_2016_05_17_15_35.png
screenshot_www_computerbase_de_2016_05_17_15_35.png
screenshot_www_computerbase_de_2016_05_17_15_36.png

base clocks reported, so add the difference in baseclocks to the stock boost and you'll get effective boost OC
 
Fixall said:
Great write up Brent! Really solid performance at 1080/1440. After reading a few other reviews, I'm super interested in seeing some overclocked vs overclocked results soon! It seems like the 1080 may start losing a little bit of it's advantage when measuring OC vs OC frame rates against the Titan X/980 Ti as it looks like the (founder's edition) 1080 is running near it's power/temp limit.



Personally... I'm waiting for either the new Titan/Ti model (or a second Titan X if the price drops enough). I purchased a GPU right on the cusp of the performance I was looking for for 1440 last generation and it left me wanting.
Click to expand...

Arstechnica has done some OC testing to 2ghz/11ghz. However their FPS gains are modest enough that I suspect a fair amount of the time the OC was thermal throttled out of existence.
 
well that performance is good, but still not 4k 60 rdy yet. I will say this card looks to be the first single gpu you can play at 4k pretty well.

Looks like I am waiting more. My 4k monitor will continue to taunt me in games at least it's good for productivity.
 
DanNeely said:
Arstechnica has done some OC testing to 2ghz/11ghz. However their FPS gains are modest enough that I suspect a fair amount of the time the OC was thermal throttled out of existence.
Click to expand...

this is exactly the same as reference titan x and 980ti reviews, achievable clocks on high-end air and water will be slightly higher but much more stable. I doubt this reference card can actually hold 2050+mhz
 
T_A said:
Ok now to sell my GTX 980 so i buy this and keep staring at the game icons on my desktop at higher FPS !

i procrastinate even for gaming!
Click to expand...

This hits home. On a daily basis I open Steam and look at my 300 games (about 250 being backlogged), only to decide... Meh, I think I'll browse [H] or Reddit. And here I am hoping the Titan X used market falls on its face so I can buy a second. >.<
 
ddarko said:
I think the conclusion is odd. The benefit of async compute is discounted because it doesn't enable a last gen 28nm card to beat the latest 16nm card? This seems to be an impossibly high bar to set. The question is not whether Fury X with async compute could match Pascal - I don't think anyone expected it would - it's whether Polaris/Vega will show similar gains with async compute that its predecessor does. And whether Pascal's reduced negative gain versus Maxwell is the result of pure brute force or architectural changes that might eventually lead to Pascal showing gains with DX12/async.
Click to expand...
We do not have access to that hardware so we cannot form an opinion.

ddarko said:
Does the theory of AMD-optimization for AoTS have any supporting evidence?
Click to expand...
None. It is an opinion, that is why the sentence starts with "I think."

ddarko said:
It's certainly true we need more working DX12 games for benchmarking but while we wait for that and unless there's evidence that challenges what the developers have stated, don't see what justifies attributing the results in Ashes of the Singularity to AMD optimization.
Click to expand...
Thanks for your opinion.

Astrodave said:
have yet to see anyone review this card for VR. VR is the only reason I would consider buying one of these now and it would be nice to see it vs a 980ti for VR. Let me know if anyone comes across a thorough review that included VR comparisons.
Click to expand...
I truly wanted to spend a lot of time with VR for it with launch, but not enough hardware to do so.
 
Makeroflostsouls said:
I'll be picking up 2 , 1070s non refrence when they come out. Should kill my dual 970s at 4k..
Click to expand...

I'm looking forward to seeing what the 1070 (Non-FE) will do, with the increase the 1080 saw, 8GB of ram, etc, I think it may even give the 980TI a run for it's money at a much lower cost, and since I game at the old forgotten 1920x1200, it should give me a nice increase over my AMD 270Xs
 
Ieldra said:
this is exactly the same as reference titan x and 980ti reviews, achievable clocks on high-end air and water will be slightly higher but much more stable. I doubt this reference card can actually hold 2050+mhz
Click to expand...
So uhh, somebody must have slapped a G10 on this bad boy, yeah?
There are no waterblocks / custom heatsinks yet but the AIO mount should fit.
 
DanNeely said:
Arstechnica has done some OC testing to 2ghz/11ghz. However their FPS gains are modest enough that I suspect a fair amount of the time the OC was thermal throttled out of existence.
Click to expand...

thelead said:
Yeah, reading a bunch of info of how well it overclocks... I'm sold.
Click to expand...

Ieldra said:
this is exactly the same as reference titan x and 980ti reviews, achievable clocks on high-end air and water will be slightly higher but much more stable. I doubt this reference card can actually hold 2050+mhz
Click to expand...
That is all worthless without actually showing me in-game boost clocks while gaming after the cooler gets heat saturated. We saw heat saturation on an open bench at ~75F ambient room temperature.
 
This really isn't enough to tide me. My 390x runs everything well at 1440p so I can wait a little longer to see the aftermarket cards now. Thanks [H] for uncovering the overclocking exaggeration. 2.4ghz on air I would have been buying one at release.
 
One thing that stood out looking at various reviews.
Seems the more recent games and those where AMD did better than Maxwell 2 have a larger performance gap between the 1080 and even the 980ti; biggest examples such as Hitman 2016 and Rise of the Tomb Raider.
From that trend perspective (and appreciate it needs more data to be proofed) it seems the 1080 is better option in terms of longevity than 980 and 980ti for overall gaming.
Cheers
 
The GeForce GTX 1080 Founders Edition is the fastest video card on the planet when it comes to today's games.
Click to expand...
Is it faster than the Radeon Pro Duo? If not, this statement should probably be amended to say it's the "fastest single-GPU video card" just for the sake of accuracy.
 
Think about all that for a minute, just yesterday the GeForce GTX 980 Ti and Radeon R9 Fury X were the best you could buy in video card performance. Now here is the GeForce GTX 1080 (with partner boards possibly cheaper and faster) providing 30% better performance in every game thrown at it.
Click to expand...


This is quite common with GPU launches, the Titan X was also 25-30% better than the 980.
NVIDIA GeForce Titan X 12 GB Review

Either way, that statement is not exactly correct. The best graphics card you could buy was an any after-market 980ti which delivered around 20% better performance than the 980ti (at 4k).
Gigabyte GTX 980 Ti XtremeGaming 6GB Review

I'm not saying that after-market 1080 won't be as good, but that statement is just not correct, the difference is about ~10%.

Other than that, performance is about as expected. the 1080 really has great perf/watt, and easily beats the previous gen after-market 980tis, it just doesn't blow them out of the water.
Maybe better cooling is where they will shine with great OC, as temps seem a bit high, and some reviewers didn't get 2ghz+ boost clocks to actually hold for long.
 
Ieldra said:
GTX 1080 OC results

screenshot_www_computerbase_de_2016_05_17_15_35.png
screenshot_www_computerbase_de_2016_05_17_15_35.png
screenshot_www_computerbase_de_2016_05_17_15_36.png

base clocks reported, so add the difference in baseclocks to the stock boost and you'll get effective boost OC
Click to expand...

Looks like an overclocked 980 Ti hangs with a stock 1080. About what I was expecting.
 
Thanks Kyle, well my rift arrives tomorrow (finally) and am still on the fence for this card since I run a 980ti now. I will probably grab one of the cheaper ones in a month or so but $699 for this one is unreasonable imo for such a minimal gain in fps unless it just stomps all over the 980ti in VR games which I have yet to see any proof of from the reviews so far.
 
So no tests were run with a custom fan profile that took it to 100% by say 70 degrees? That might have made quite a difference on being able to maintain a higher boost clock.
 
Now that we have solid numbers on the 1080, can we estimate the 1070 will be 10% (average) faster then the 980ti?
 
DX12 performance is all over the board.
GamersNexus got crazy good numbers from both Maxwell and Pascal.

NVIDIA GeForce GTX 1080 Founders Edition Review & Benchmark

Asynchronous compute, however, sees major gains. Some of our titles reported a frametime performance improvement from Dx11 to Dx12 approaching 50%, where previous cards (like the GTX 980) struggled to reach even 5% improvement (Dx11->Dx12).
Click to expand...
 
Gatecrasher3000 said:
Now that we have solid numbers on the 1080, can we estimate the 1070 will be 10% (average) faster then the 980ti?
Click to expand...

I'm expecting the 1070 to trade blows with the 980ti after overclocking the snot out of both. Looking like the 970/780 Ti situation again (which is damn good for consumers imo).
 
CSI_PC said:
That is 21%-30% across the board with beta drivers....
It will only increase with drivers done like they are for the 980/980ti.

Still not worth upgrading if still own a 980ti and happy with current gaming and can wait, but that is a good performance jump for the replacement to the 980.
Cheers
Click to expand...

I think it's just because those games are where compute throughput was most relevant to performance, and the 1080 finally takes back the crown in stock compute performance.
 
This new card is dope! Pascal is going to be knocking our socks off for a while!

When can we learn more about the 'fast sync' type of technologies? Not sure what type of monitor to pair this with, my 60hz IPS 25x14 display may not be able to show the performance in all its glory (especially during FPS's).....
 
Last edited:
Wow! I am really impressed. Handily beating the 980 Ti by that much for the same money and less power and heat? That seems like a no brainer really especially if the AIB's start coming out with them at $599. I dont really need a GPU upgrade right now with my 290x @ 1080 but its nice to know we've finally got some worthwhile cards to upgrade to and these little single digit performance increases the last couple go-rounds looks to be over.
 
Very nice work guys -- I was hoping for 25% over 980ti... and seems we get 30%.

For someone who's been using a 970m for the past 2 months waiting to fill the gaping hole in my desktop gaming rig, seems that selling my Xfire 290's a while back was a smart move. Will be more than happy to drop 600-650 on a nice third party card, my personally favor anything by MSI. (not a fan of blower/reference styles)

This is just the beginning too -- will be exciting to see what binned chips from other companies will crank out with custom cooling and whatnot. Not to mention driver improvements :)
 
Thanks [H] for all the effort in preparing this review. This article was all I needed to tell me that this will be my next card. Now to save up.

I hope we'll see some 3rd party card with nice clock speed boost (MSI Lightning, Asus Strix, etc)
 
spintroniX said:
I will probably be picking one of these up later in the summer. It's a monstrous upgrade from my 780.

I just want to see what the price to performance of the 1070 is like and/or what AMD has on offer
Click to expand...

Same here, my 780 struggles hard on Doom at 1440p, getting between 30-50 fps.
 
Pretty impressive, but the blower continues to disappoint. I wouldn't be surprised if it could hold that 1885 MHz Boost if the fan curve was set to go up to 100%. All this makes me excited to see what the AIB cards will do, and even more excited for GP100.
 
Creig said:
Is it faster than the Radeon Pro Duo? If not, this statement should probably be amended to say it's the "fastest single-GPU video card" just for the sake of accuracy.
Click to expand...
But a Radeon Pro Duo is NOT a gaming card, is is a professional content creation card. When we asked for one repeatedly, we were told by AMD that it is NOT a gaming card, and AMD refused us a sample. And quite frankly, I do not spend our resources on content cards for game testing.

thelead said:
Good to know. Any info from manufacturers on when we will start seeing custom 1080's?
Click to expand...
Hopefully by the end of the month.

ddarko said:
A glib reply to a serious point. Because it's an "opinion" isn't a free pass to say anything. Not all opinions are the same; the opinion that "I think blue is the best color" is not open to dispute but the opinions that "I think the sky is the color blue" or "I think a game has been optimized for a vendor's hardware by the developers" are definitely subject to factual support or refutation. If you want to ignore that distinction, there's nothing I can do except point out the absence of any evidence supporting your "opinion."
Click to expand...
Your thoughts are noted.
 
No OC numbers? Looks like I'll be waiting for that review too. Thanks for this though!
 
You must log in or register to reply here.
Back
Top