jlp209

stalinx20
Has anybody tried underclocking the card yet? 

Yes I tried this last night w/ my 1080 FTW Hybrid after checking out that old thread, interesting. Moved the core offset and memory offset to -200 in Precision. Ran Heaven just fine for 15 mins. Still have yet to experience the BS100% with this latest card. 

If more people can concur and confirm their card is stable with the offset, then we're right back to the same old song that the FTW is overclocked too high. 

HeroicArchetype
I am on my 2nd RMA now, not going through advanced since I am off Europe for a week or so starting by next week. I hope everything will be fine.
 
This crash happened yesterday because I forgot to start the custom fan speed with msi afterburner

If I understood that right you are saying that the card works OK only if you use a custom fan setting?
 
That's pretty much the conclusion I've come to with my current "defective" card. I found that by enabling the custom fan profile in MSI Afterburner (40% baseline then roughly 1% for each 1 C increase above 40 C) that the black screen issue has not happened for the last four days. So does that mean that the problem lies with the default fan profile which only turns the fans on once the GPU hits 65 C? I've never noticed any issue with my own card ramping up the fans in games using the default fan profile; it always came on above 65 C and I never saw temperatures higher than 81 C during the summer with temps generally being in the 65-77 C range for the majority of games (all v-synced on a 60 Hz display).
 
If that is the case then it makes me wonder if this issue isn't more widespread but people who are using custom fan profiles aren't experiencing black screens. If the fans are off then that means the GPU is running cool but what about the VRAM or VRMs? Could the issue be due to those components getting too hot due to the fans not spinning such that when a game is run they end up overheating after 5 minutes or a few hours?
 
The FTW card has a dual BIOS with the second one offering a 30% power increase to 130% and a fan profile that spins all the time. Is this BIOS free of the black screen issue? Has anyone tested it? 

We already know that debug mode in the nvidia panel stops the crashes, which is basically just an underclock. For me, it locks the card at 1970mhz or so, which is still very fast, and I get no black screen then. However, my replacement comes tomorrow so we shall see how that runs. Obviously, you buy these cards expecting to overclock and if its unstable at stock, its not stable in overclocking.
Also, switching BIOS makes no difference. Crashes galore.

jaju123
We already know that debug mode in the nvidia panel stops the crashes, which is basically just an underclock. For me, it locks the card at 1970mhz or so, which is still very fast, and I get no black screen then. However, my replacement comes tomorrow so we shall see how that runs. Obviously, you buy these cards expecting to overclock and if its unstable at stock, its not stable in overclocking.
Also, switching BIOS makes no difference. Crashes galore.

That is not what my point was - at all, but to give EVGA insite that their previous issue with the 980FTW (yes, it was that big) has the same relevance to the issues of the 1080FTW. They are clocking it too high. The statement someone made about an EVGA rep claiming that the issue has been fixed was the same comment made when they "fixed the 980FTW", but the issues still remained. The fact that you all are able to use the card at "debug mode" only proves my theory, which is the same theory i had for the 980FTW. Then, some of you complain that you get your RMA'd card but the RMA still has the same problem. Good times again. Same old song, same story mirroring the 980FTW. Which was the main reason why i included the thread concerning the 980FTW. Carry on. 

I believe the debug mode just alleviates the problem, I found it interesting how a post on 1st or 2nd page in this thread fixes the issues by putting a fan that blew air across the top of the card.
 
Unless they decide to actually tell us what is wrong I'm going to believe something on the backplate gets hot and unless you have fans blowing at it (cpu cooler, intake, outtake) then it's going to black screen more easy. AIO coolers probably make it worse since they generally don't have the airflow over there.

stalinx20

jaju123
We already know that debug mode in the nvidia panel stops the crashes, which is basically just an underclock. For me, it locks the card at 1970mhz or so, which is still very fast, and I get no black screen then. However, my replacement comes tomorrow so we shall see how that runs. Obviously, you buy these cards expecting to overclock and if its unstable at stock, its not stable in overclocking.
Also, switching BIOS makes no difference. Crashes galore.

That is not what my point was - at all, but to give EVGA insite that their previous issue with the 980FTW (yes, it was that big) has the same relevance to the issues of the 1080FTW. They are clocking it too high. The statement someone made about an EVGA rep claiming that the issue has been fixed was the same comment made when they "fixed the 980FTW", but the issues still remained. The fact that you all are able to use the card at "debug mode" only proves my theory, which is the same theory i had for the 980FTW. Then, some of you complain that you get your RMA'd card but the RMA still has the same problem. Good times again. Same old song, same story mirroring the 980FTW. Which was the main reason why i included the thread concerning the 980FTW. Carry on. 

Of course I get your point, and I was agreeing with you - underclocking via debug mode fixes the problem. I do, however, disagree that these cards are being clocked 'too high'. Why do I disagree? Well, because they are not clocked any higher than the competition, and the competition do not have this problem. How do these cards differ from the rest of the pack? The custom PCB. Thus, I think that some aspect of the custom PCB design caused this flaw, perhaps power delivery for example, whereby the issue becomes worse the higher the card clocks.

 
The flaw with these cards is the same flaw they had with the 980FTW. the competition design their boards differently than EVGA's, and the way EVGA has their PCB made, since they're running at the same speeds, apparently there is something wrong. We can agree to disagree, but i think you missed the part that this is the same fluke that EVGA had with the 980FTW. The exact same issue. 

stalinx20
 
The flaw with these cards is the same flaw they had with the 980FTW. the competition design their boards differently than EVGA's, and the way EVGA has their PCB made, since they're running at the same speeds, apparently there is something wrong. We can agree to disagree, but i think you missed the part that this is the same fluke that EVGA had with the 980FTW. The exact same issue. 

Yeah I get that too, and it's obviously kind of inexcusable. It may be possible they literally don't know what is wrong in their design lol, so they just keep trying different components from different sources.

http://forums.evga.com/badHOT-PCB-Layout-on-GTX-1070-FTW-m2565921.aspx
This could be the explanation!

NetQvist
I believe the debug mode just alleviates the problem, I found it interesting how a post on 1st or 2nd page in this thread fixes the issues by putting a fan that blew air across the top of the card.
 
Unless they decide to actually tell us what is wrong I'm going to believe something on the backplate gets hot and unless you have fans blowing at it (cpu cooler, intake, outtake) then it's going to black screen more easy. AIO coolers probably make it worse since they generally don't have the airflow over there.

That might explain why using a custom fan profile seems to prevent (or reduce since I am at four days and counting...?) the black screen issue on my PC since the air would be cooling not just the GPU but also the other components. The default profile has the fans only spin when the GPU hits 65 C so until then the VRMs and VRAM are not getting any cooling additional cooling, only whatever airflow your own case may have.
 
In my case, I got the card is early August when it was warm in the UK. Initially, I was somewhat disappointed in the ACX 3.0 cooling as I was seeing temperatures as high as 80 C in some games and so opted to switch to a more aggressive custom fan profile using MSI Afterburner. I used for over a month until I switched back to the default fan (the one that only spins at 65 C+) about 3 or 4 weeks ago. The black screen issue started happening for me around two weeks ago. Make of that what you will but it seems to me that the problem may well be with PCB components getting too hot due to inadequate cooling, especially when the card's own fans are not spinning (because the GPU has not reached 65 C).
 
Then again I accept I might be completely wrong... lol