Have you tried to disable full screen optimization to see if it makes any difference?
 

 
Are you using displayport cable? If yes, is it vesa certified?
 
XFX 750w is new?
 
GPU temps were ok when the crash occurred?

Geoshot
how would I check for vesa certified?

#1 See if the cable carries the dp logo on it.
 

 
Only products that undergo testing are permitted to carry the DP logo. Companies that use the DisplayPort logo must also be a VESA member.
 
#2 See if your cable is listed in the database listed here.
 
 

I just checked it, it's actually a Display port to HDMI cable - DP from the GPU -> HDMI to the monitor
The display port part does carry the logo, it's a Benfei DP Display Port to HDMI cable - But I cannot see this manufacturer in the cable database.
 
No the GPU fans are still sitting at the normal speeds they were when the game was running, bare in mind this is a hybrid watercooled GPU so the fans never speed up much at all.

Ok, so I'm going to test this more later.
 
First I will run just with a single monitor - HDMI > HDMI 
If it crashes, I'll go back to the duel monitor setup and disable full screen optimizations, though wouldn't this mean I would have to do this for ever game I play? Seems a bit odd that this would randomly become a fix after the card has been working for so long.

So I've been running a single display HMDI > HDMI for about 2 hours playing The Witcher 3 (Which is the game that would crash the fastest) 
and thus far I've not had a single related crash, there was one CTD while fast traveling in game, I'm putting this down to a bug only and not related.
 
**Update**
 
I have tested another hour or so on Escape from Tarkov - Didn't run into any issues. I've now reconnected the secondary screen both via HDMI. and will do some further testing in the evening.

I got another crash while streaming, nothing unusual going on just queuing for the next raid in Escape From Tarkov, and the whole system reboots all of a sudden.
 
The only error I get in event log is the standard system lost power

"The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly."
 
But like I said before, I've tested every other component in my system it has to be the GPU.
Honestly, it's looking like an RMA job. :(

Yeah another crash just now again, it seems the timing of when crashes happens has absolutely no pattern or consistency. 
I'm going to pop in my older R9 280x tomorrow and see if I can play some single player games during the day without any crashes, and stream and night and test the same. if I don't get a single crash it's 100% an RMA. 
 
If I get a crash... let the investigation continue! 

Dr.Death
I can give you some things to  look at on my 980ti hybrid ??  not to say its the best  but maybe give you some guide  to look over  or at  that may help [at box stock]
 
hwinfo64  results after firestrike stress test
 
 
after 3ed loop during 4th  of valley and gpu-z render test  running at the same time
 
 
maybe see if your  psu is dropping out under a good load ?
 
 
 
 
in my opinion  I think this is way too much orange in the blue field   [notice how clean blue my card stays  ]  some small here and there spikes  is fine  [ of course that's my opinion on that]
 
[from a web site review ]
 
 
 
maybe a bad or going bad card ?   then it still can be anything in your system ??    be sure you did a good clean driver uninstall  and maybe try another driver next  ? I still use older like 355.82 with my 980ti hybrid    [ I get too many issues like you got with newer ones ] right now I using  353.62   due to this  [matter of fact  it was not intill aftyer I tried a later driver that my black screen issue started   [like around driver 372 ???? ]  befor it I never had it  occur  
 
 
 
 
 
so you can see  that that game worked well under  353.62  but the next newer 355.82  NVidia broke that game  in it .  so its easy to be a driver  with NVidia now a days     . just because its a new driver don't make it a better driver  in any way . just go to try some out  . and I stick with drivers made at the height of the 900 series  releases . now there focus is to 10 series stuff / support  [opinion]
 
like one guy said  '' seems NVidia brakes as much as they fix  in each new driver ''  LOL....... so true 
 
can you test your card in another like system  ? maybe you got a buddy with a nice rig to try it in ?
 
in the end  if you feel a RMA is the best  that's all you can do and try  , just don't look to get a hybrid 900 series back  . maybe a 1070 air ??
 
 
 
 

Thanks for that info, I'll try some of the extra tests maybe.
Right now I have my older R9 280x in my system, and I've been able to play for several hours and not get any crashes at all. 
While playing last night the 980ti crashed after 43 minutes, then a second crash after 123 minutes.
So far I've played over 240+ minutes on the AMD card and it hasn't crashed the system at all.
 
One really important thing to note about the drivers, is I haven't lost any performance at all on the card. No issues at all with the games themselves, it's just the hard crash and reboot comes randomly.
 
I've submitted an RMA for my card, but after tonight's stream using the R9 280x I may put back in the 980ti and revert back to an older driver just to test and see if the crashes stop. I would rather keep my Hybrid 980ti than get 1070 air, the performance difference is tiny, but the temperature difference is huge.
If I was to get a 1070ti or 1080 though, then I'd be happy about doing the RMA haha. I've seen some people get 1080's from a 980ti RMA.
 
This is also the second PSU I've tested on this, I also have a spare i5 4570 to test as well, but I'm 99% sure it's not the CPU as I've performed hours of Prime95 stress testing and CPU intensive tasks like Streaming and Rendering and I've never had it crash, only while playing games and that's a GPU task.
 
I'll update back when I have some more details regarding diagnostics with the driver roll back, or if anything changes with RMA status or if the rig crashes with the AMD card. (It's not going to though.)
 
 

Dave3d
I can save you the trouble:
Its your systems fault.
All the time.
Every time.
And when you have tested everything, and re-installed the os, and everything works without the 980 in there, and you put in the 980: You guessed it, it will still crash, and it will still be your systems fault.
I hope you dont get the infinite 'its your systems fault, your card tested fine' like I have gotten, and you get garbage cards back in return.
 
I AM hoping it is your hdmi port or cables though, since it seems to work fine in a single monitor cfg, but not a dual monitor cfg?

Regardless of single or duel monitor cfg the system crashes.
The system crashes if the 980ti is the graphics driver.
The system does not crash when using my older AMD R9 280x.
Crashes happen randomly while playing games, there is no timing trend to them.
Crashes completely reboot the machine without warning, no errors or log in Event viewer.
All cables has been replaced with brand-new ones, Cables have been tested on both GPU's - No crashes on AMD card.
My AMD card draws more power than my 980ti - If this was a PSU related issue then switching to the AMD card would only make it worse.
 
I'm going to continue to use the R9 280x for the next 24 hours, if i do not get any crashes then it's safe to assume the issue ONLY effects the 980ti.
I will then try some more diagnostics on the 980ti, such as driver roll back. Though I am 99% sure this is a hardware issue and not a software one, as a software issue would give some kind of error or log to read.
 
My RMA has been accepted, so if no solution if found before the weekend, I will be sending the card back to EVGA for replacement - And prey for a decent 1070 or better.

Dave3d
Ah, I thought you had said the system didnt crash with another card, but didnt see that it was an amd card.
Have you tried different drivers?
 
I have had the same thing happening to my system for over a year now (well, that ebga knows about. I put up with the micro stuttering and losing framerate issues for a long time before rma'ing) (it started with micro-stutters, then desktop took minutes to load, then I couldnt even get to the login screen (1st card), and the rest have had all crashes, DP ports not working, framerates that go regular then suddenly drop to insanely stupid levels like 20 (in games where I have played years, and the frames never go below 70's), and I just keep getting sent worse cards than the previous one.
And, they tell me that all of the cards are good, they tested good.
But, they dont seem to game with the cards (and some dont even seem to get tested at all before being sent out), and its the games (as well as 3dmark FireStrike tests) that crash, or fail.
 
I will say one thing though, everyone thought that my issue was a psu issue, and I did see a very low 3.3v in the bios (random chance I saw it, I wasnt looking for it), and I got another psu to put in the system (an OLD 1200w), and the crashing stopped.
So, dont rule out a psu as the issue.
Mine was an EVGA 1300w Gold G2, only a couple years old, so it does happen that it 'could' be a psu issue.

You might have missed it in previous posts, but This is the second PSU i have used while still the crashing persists. Originally i had a Corsair 600w, then i swapped it out for a XFX pro 750w.
I don't have the money nor the ability to buy/borrow a third one so I can't check if this psu is faulty too, but tbh it should be completely fine. The AMD R9 280x draws more power than the 980ti does, so if it was a PSU issue i would get crashing with this card also.

Dave3d
Yeah, 2 psu's and still happening, ouch.
 
Just a couple Q's, for my own info:
Do you have an M.2 card for the os? Or is that a regular sata ssd?
Is either of those monitors G-Sync/ULMB/Free-Sync (like maybe the 21:9?) ?

No M.2 just a regular Samsung 850 Evo over Sata 3.
The AOC 27" is freesync compatible over Display port 1.2 - However with Nvidia cards i've never been able to utilize it before, until now actually while im using the AMD card.
The 21:9 monitor isn't G/Freeysnc and I don't know what ULMB is, but i just use it for production/streaming windows & watching true 21:9 films without black bars :D

i do know  Witcher 3 comes up with complaints in a lot of NVidia driver feed back threads .. I got crisis  [first original one]  and it runs and plays  fine . [matter of fact a bit better with my old AMD card]
 
go figure ???
 
one thing you can look up is with 900 cards  there was some issue with the ports  [display, hdmi,dvi ]   gigabyte cards  seemed to have the most failing ones reported  but it seems a overall 900 series issue  with any brand ?  that goes well along the lines of using dual monitors  .. a bit of gooling it up  may show you some on that  ..
 
also seems like guys with higher end cards like the  classy or kingpins get offered like a 1080  -  the rest of the line is next best card like a ref 1070 ??
 
just discuss this with your evga RMA support guy  when that time comes  .  may come down to what ever is handy to send out  ???
 
[claims a 1080 for a hybrid 980ti ] [also a like issue as you ]
  https://forums.evga.com/PC-is-Crashing-screens-go-blank-suspect-980ti-hybrid-m2595004.aspx
 
https://forums.evga.com/What-card-did-you-get-after-RMA-980Ti-Classified-m2731794.aspx
 
don't know I assume if they had any referbed 980ti hybrids ready it would be a more direct replacement as well ?????   luck of the draw ?????

could be ? you'd think that 290x is a more power hungry / demanding card in full load use  , so to me I would almost feel  its not now a power issue like from the board or psu [???]
 
still cant say much on NVidia drivers on a lot of cards behavior  [???]
 
unless you can test that 980ti in another known good system  to see if it exhibits  the same behavior  as in yours then that could nail it to be the card for sure  [???]
 
I guess if not but you feel its a RMA  then roll the dice on that  . the replacement will work as you expect and good or it don't and your in the same boat  .
one thing I do know is I do have to juggle drivers for my NVidia cards  .  I had amd as you  do  but there drivers just drop support for some games and programs I use 
 
like  some older games if you use drivers over 12.6  they no longer work  and 13.12  was the last latest driver I cud safely use  with out more issues   [you seen the black screen issues with them 200 series cards as well after 14.xx ]  I seen that as well  under my AMD card   with that 13.12 I never had any of them issues  out side of needing to use a older driver as 12.6 for some of my older games . also I got some games that need older agea physix   that's not supported under AMD at all  , but work fine under NVidia cards with the old agea PhysX  support drivers
 
so I went NVidia this go round and looks like I did not improve much   all my games and programs work  , but ?????
 
so far lucky my black screen is just at start up   and only on this hard drive with full fledged  win 7   never happens under the 2 Linux drives and the drive with raw win 7    [useing the same drivers and all   just this win 7 is not activated or ever been online or updated ]
 
one thing you got to know is now a days   all this hardware  is based on hype, gimmicks   quality and support is way in the back seat    its now led lighting and logos  stupid dust covers / dust collectors . are more important selling points today  [sad but true]
 
good luck
 
 
 
 
 
 

ya,  I guess  , you can only do what you feel is best  for you.  you can only struggle so much  then try a change .    best of luck
 
post back on what you get on your rma replacement card  and then when you get it how it worked out or resolved

Yay!
I hope you get a good one.
 
I am on my 6th rma, and hoping this one will be good.
Mine should arrive tomorrow