I've had a couple hrs to sit with my 1080. I have it +85 +300 with 120% power stock voltages temps dont pass 67c.
Its stable with fire strikes stress test however soon as I tick +90 on the clock it fails and fails hard.
Does any one have some experience increasing the voltage on a 1080 like how much is acceptable and safe etc.
Im not looking for alot but a even +100 would make my day.

orlando2542
RMA it that is totally no bueno
 
I have stock voltage 
+170 GPU 
+500 Memory 
70 C  

 
Overclocking potential over what is advertised as the clock speed is not a reason to submit for warranty replacement.
 

tubbing
I've had a couple hrs to sit with my 1080. I have it +85 +300 with 120% power stock voltages temps dont pass 67c.
Its stable with fire strikes stress test however soon as I tick +90 on the clock it fails and fails hard.
Does any one have some experience increasing the voltage on a 1080 like how much is acceptable and safe etc.
Im not looking for alot but a even +100 would make my day.

 
Under load with your stable over clock what are the boost clocks you are reaching?

During benching it runs at 2025 solid during firestrike stress test. But soon as I move the slider to anything plus 85 it fails regardless how much voltage I throw at it.

great how companies advertise with "made for overclocking" knowing that it's just a big lie.
it's always the same exactly like Intel's XMP, "it's there, buy it but no warranty it will work for you"...

So far settled at +100 to core and +350 on memory with stock voltage.  Might could get a little more but it already boost to nearly 2.1 and with over +150 on the core I was getting some artifacts.

Lokius81
Even on stock settings it seems like my 1080 SC runs quite hot..... i have to move to fan profile to 70% manualy just to keep it in the low to mid 70s.... is this normal?
Its quite noisy at 70%....

 
Was the card reaching thermal throttling temps before you adjusted the fan? If so than yes, that's bad. If not than it's fine. A card seeming kind of warm by you while running at 70% fan is clearly not a problem with the card. Also... 70% fan? They made 100% fan where it is to keep the GPU cool enough to operate normally. If you can get away with only 70% then congratulations!

I'm with sethleigh. I do not understand the fascination with firestrike scores or highest overclock. The thing was made to play games. And like he said: the difference between stock FPS and overclocked FPS is 1-2%

zshadez
 
It's funny how in trying to be reasonable you're generally insulting instead. You don't know the setup in question, so defaulting to "e-peen" comments is childish and helps no one.
 
My setup is either to 3440x1440 or 4k, in either case a good overclock on the 1070 is generally the difference between choppy and smooth framerates. Gameplay is noticeably much better with the OC, I'm finally stable above 60 on my primary monitor.

The OP reported a +85 MHz overclock. His e-peen better reported +170, and suggested that the OP RMA his card. Think about that. EVGA sells a card as factory overclocked by over 100 MHz as compared to the base model. And the card clearly achieves the advertised speed stably. In fact, he achieves that and an additional 85 MHz. That's pretty good value for money, and EVGA's card performs better than advertized. So, what's the justification for an RMA? You do realize that with an RMA EVGA is going to lose money, right?

But consider the difference between an 85MHz overclock and a 170 MHz overclock. Starting at 1708 base clock for the out-of-box SC, add 85 for the OP and 170 for the guy who thinks the OP is entitled to make EVGA lose money on a card that works better than advertized. That's 1793 MHz as opposed to 1878 MHz. Dividing those we see about a 4 1/2% difference. If the guy with the longer e-peen is getting 60 fps on his 4k monitor the OP is getting 57 or 58. In practice you're never going to see that 2 fps difference.

And that's assuming that the boost clocks are even all that different. I don't recall seeing them compare actual achieved boost clocks under load. We're just assuming that the guy with the higher offset to the base clock is actually getting a 4 1/2% advantage, but I haven't seen the proof that he really is.

Trust me about 4K, I feel it too. I'm running a 4K monitor with a Geforce 960 SSC right now, and feeling the pain. After playing The Division for a week at 20-30 fps I've bit the bullet and down-rezzed to 1080p just so I can peg it at 60 fps for playability. I'm trying to buy a 1080 SC as well, and have been glued to the websites for several days now trying to catch one in stock. I appreciate the advantages of buying a faster card. That's why I'm trying to buy one, like so many others.

My whole point is that with so many of these guys it seems as if the most important thing to them when they first pull it out of the box is just how much further than stock they can push their overclock, as if the overclock itself were the end, and not just a means to an end. If the true end is more playable quality and framerates in games, then a little common sense kicks in. If my overclock ends up being 50 MHz lower than the next guy's, then yeah, I suppose he wins the e-peen contest, but all other things being equal, he might see 1 or at most 2 fps advantage over me in any games we might play. That's just not a big deal, and certainly not worth losing sleep over. With these things the law of diminishing returns applies. What are you willing to do for that last 20 or 30 MHz of overclock? And what if you achieve that last 20 or 30 MHz of overclock? What real world performance actually using the card have you gained?

Point taken, and I appreciate your comments. Probably not the best tack to take in conversations like this, and I'll reevaluate my approach for the future. I think the whole e-peen thing helps people see how ridiculous this whole thing is though.

I've overclocked off and on for like 25 or 30 years in machines I bought or built. I get it. I've always used overclocking as  a way to eek out some better performance if the benefit were tangible enough. It seems nowadays though overclocking is like a test of virtual manhood, where the contest is to see who overclocks the best/has the luckiest draw in the silicon lottery. The overclock itself has become the end, rather than the means to an end, which is more pleasurable use of the machine. I am not a fan of this newer focus on the overclock itself.

It's why I'm still running a 5 year old CPU in my machine, btw. Until 9 or 10 months ago I was running a Core i7 950 quad-core at 3 GHz. I found a Xeon 5680 6-core 3.3 GHz for cheap (maybe $125) and dropped it in for a nice upgrade. Is it as fast as today's cpus? No, certainly not. But I'd never notice, because in Windows itself, and in almost any game I play, the cpu isn't that big a deal - it's all about the graphics card. It's why I'm looking to upgrade to a 1080 SC card to run my 4K monitor better, but I still am perfectly satisfied with my 6-core Xeon from 2011. And it's not even overclocked at all.
 
I'm currently doing a little scouting to possibly move toward rebuilding my machine with an X99 board and an Intel 6850 at some point, but not because I feel I really need the extra oomph, but just because at some point the caps on my X58 board will go bad and my machine will become unstable, and I'd like to make a move up to a new generation before I get to that point so I don't have an unexpected failure and be out of my machine for a while.

No doubt, well that's not my thing with computers, but I totally get it. I've built two violins and an electric guitar (and tube amp) from scratch, and I don't actually play either violin or guitar! I like doing things too, so I totally get that.
With computers I'm an unrepentant gamer though. I tinker with my machines too sometimes, but I really do play games, and way too much of it. I actually spent a couple of evenings a few months back overclocking my new Intel Xeon 5680 cpu to see what I could do it with it. After all that I set it back to stock settings for actual use, because I didn't think that the performance gains of aggressive overclocking were going to get me anything that I'd actually notice in real usage situations, and mild overclocks even less so. So why stress the equipment?

The sad thing is I like tinkering enough that I put in a Swiftech 220X water cooling system in my machine, and I'm running it stock! Talk about totally unnecessary. Oh well. I'm well set up, however, if I want to watercool whatever 1080 card I end up getting, or if I do move up to an Intel i7 6850 I can water cool it from day 1.
 
I suppose if people want to spent time and money building machines just so they can measure them and compare measurements with other people on the internet, I can't really find fault with that. I've got two violins and an electric guitar I can't really play well that would make me a hypocrit if I said otherwise. Still, I can't resist the impulse to remind some of these guys that their computers are actually made to be used, too. Stroking their e-peen is all fine and good if that's what they really want to do, but at least they should man up and admit that's what they're doing.

EpicDonkey
As far as having the fastest OC, its like tuning a car to make it the faster.  Some people just to do it to see how fast they can make it go.  The 1080 can easily play just about anything at 4k with decent frame rates.  You might have to tinker with some settings but more often than not the difference will be negligible.

I'd say a 22-27% improvement in core speed is more than a negligible difference. At 3440x1440 it's taken multiple games from ~50fps on ultimate to completely steady 60 with vsync.

The_Ether_Bunny
 
He is saying, I assume, that the difference between the card's boost clock and max overclock is negligible. Which it is. When the card is running 3D applications it is not at "core" speed. The FTW automatically boosts to 1987 Mhz; max overclock to 2045 Mhz. About 2.9% increase which will net maybe 1 FPS @ 4k.

 
Results depend on the individual card and can be far from negligible. 1070 FE boost is 1683, 1070 FTW boost 1797, my FE is running at 2100-2150. That's still about 25% over stock and 17-20% over stock FTW. As I've said before, it's the difference between ~50fps and solid synced 60 for me at 3440x1440.

zshadez
 
Results depend on the individual card and can be far from negligible. 1070 FE boost is 1683, 1070 FTW boost 1797, my FE is running at 2100-2150. That's still about 25% over stock and 17-20% over stock FTW. As I've said before, it's the difference between ~50fps and solid synced 60 for me at 3440x1440.

1683 is the advertised boost. The actual boost is 2000. You can overclock it to 2150. Which is an increase of 7.5% which is not scalable with FPS. 1-2 FPS max.

My point was not that OCing a card produces a negligible difference, there are so many variables and results may vary lol. What I was saying is the performance increase you get by OCing your card can often be achieved by a small amount of tinkering in settings. For example, antialiasing at 4K can be turned down if not off and the visual difference be negligible but a noticeable performance gain.

My personal experience, I do get a bit smoother gameplay in more demanding games with my OC but I was already getting solid 60 fps with the SC's stock clocks. My 1080 SC with stock clocks has handled everything I've thrown at it as well @1440p and @4k with small adjustments.

tubbing
LOL I would never RMA a card cuse it goes to +85 instead of +150.
I run a 1440p monitor and this cards eats everything i throw at it at 1440p with DSR set to x4

x4 DSR on top of 1440? that is 5k. lol you do not even have a clue what you are saying as no way in hell a 1080 can play all games at freaking 5k with decent performance.