I'm having the same issue. Happened a couple times with Witcher 3 and just occurred tonight with Gauntlet. I don't think it's a stress/heat issue. I had HW monitor running with Witcher 3 after the first crash and CPU and GPU were not overheating before crash occurred. Using two monitors, one 27" using display port at 2560x1440 @ 144Hz, and another 27' using HDMI at 1920x1080 @ 59Hz. Both screens would go black and monitors would indicate no signal detected. System still running in background and any commands I do I can still hear working in game. Reset will cycle through the boot process but still no display. Only way to fix is to do a hard power cycle. I just got this GTX 1080 FTW less than a week ago. 
 
Edit:
I found other threads with similar problems with 9 series video cards from EVGA. Is this a common problem with EVGA? I had no issues with my previous MSI 980 that was previously installed in the same system.

My 1080 FTW looks like it's overclocking even when I have everything set to default in Precision X OC.... I think it's a software issue. I hit default, boot up a game, then the hardware monitor says i'm getting a 2k + clock speed.
 
Then no surprise, I get the same issue. I used to have this issue with my gtx 770s, I assumed it was just a too heavy of overclock.

That is just how gpu boost 3.0 works. If the problem was the card boosting too high with low temps it should artifact or driver crash i would have thought. Isnt the black screen fans at 100% a fail safe for when the card is overheating?

fakeyfakerson
Seems like it happening in World of Warcraft is pretty common here, but it's happening in other games as well.

It keeps happening to me in Guild Wars 2, especially in instanced pvp

This only happened to me on my 970 when I had low power input to the card...stopped happening when I fixed that.
 

Hi Brutal!
 
I'm having the same problem as you (coil whine, black screen monitors, fans to 100% after black screen) with my 1080 SC ACX, my problem becomes chronic when I try to play any game with nvidia DSR On.
You had success with the RMA?

o/
Well to start off I never had the coil while issue with my card.
 
I did receive the new card from RMA today, and installed it when I got home from work.  
 
Started off with a Heaven benchmark to sort of warm it up.  The results were slightly better than the first card but hardly worth mentioning.  After that I spent a couple hours playing Battleborn because that was the game that triggered crashes most often for me before.  Took a break to cook/eat dinner then played a couple more hours.  No crashes yet!
 
Now with the old card I did manage to have one night of gaming where it didn't crash.  So I'm taking this first evening with a grain of salt.  I'll be keeping an eye on things until the card has more than a few hours of lightweight gaming under it's belt.  Doom, Witcher 3 and Just Cause 3 will all make the 1080 actually try.  I've also been wanting to spend some time in Fallout 4, Dying Light and some other old ones to enjoy what it's like with everything maxxed.     

Well, I was crashing pretty consistently in Heroes of the Storm, but I noticed that the back plate of the card was (excuse the cliche) hot enough to cook an egg on. I think that something else on the PCB, maybe the ram or some other component is overheating. I ran the fans at 100% with the case open and my gpu didn't exceed 43 degrees, and didn't crash once. Before I was crashing when I hit 70 degrees and maintained that for more than a couple minutes. I'm wondering if this is a design flaw, or a flaw in just my card?
 
Can anyone else having this problem try that solution? Granted, I did a lot of other fixes before this, including step 1 & 2 in Sajin's troubleshooting guide (very nice by the way, thanks), but maybe someone can confirm.

I'm also having the same issue sadly, my 1080ftw isnt even stable if it's underclocked slightly. I've tried everything in the troubleshooting guide, and it seemed to be stable for about 3 hours of witcher 3 in debug mode. That being said I'm still going to RMA it most likely. The card should at least be stable at its intended clock speeds.