Hi All,
 
Wanted to see if anyone can help diagnose my game freezing issue.  When playing battlefield 1, I can't play 2 conquest games (~60 minutes) without the game freezing up.  While in a game, screen will freeze.  Audio goes out, doesn't go silent, but will have constant negative feedback, like a constant buzz sound.  PC doesn't auto restart, I have to hold the power button to shut it off and restart.  On occasion, after it freezes, I am able to minimize the game and there is a windows prompt from the game stating that there is a problem and program needs to close.  Event viewer doesn't show sign of bsod that i can tell, just the critical error that the system shut down unexpectedly (probably from me force powering down).
 
I do have a relatively a new set up, recently picked up a 6850k and x99 classified.  I've been downclocking a combination of my core multiplier, cache clock, ram clock, and playing with voltages for the last couple weeks, and can't gaming stable.  However, all my over clock settings have passed prime95 (v26.6 and v28) for 2-12 hours, HCI memtest (500%+), and realbench (4hours).  I've also played with the in game settings, with/without vsync, anti aliasig, memory restriction, etc.  To be honest, not sure if it's my overclock, the video cards, or game itself.
 
Any suggestions are appreciated.
 
Setup:
6850k, water cooled, 4.2-4.4ghz, tried all clocks after passing stress tests, temps <80C
980ti classified SLI, water cooled, tried stock clocks and OC'd, temps <60C
X99 classified, tried ring ratio 28-34, ram at 3000-3200, all passed 500% HCI memtest
Ram are corsair dominators 3200 32gb (8x4)
corsair 1200i psu
 
Thanks

Sajin
#1 Run the cpu @ default clocks.
#2 Run your ram at 2400MHz with correctly set timings & voltage.
#3 Run the game with sli disabled.
#4 Use msi afterburner to underclock your gpu to 980 ti reference speeds. GPU clock should read 1075 MHz or less while in game.
 
See how that works. 

Sajin,
 
Thanks for the tips.  I went straight to #4 just out of a hunch that my stress testing on my cpu/uncore/ram clock was solid.  I could not get msi afterburner or precision to go down all the way to the reference speed, but i did downclock the 980ti classifies as much as it allowed me to.  Out of the box, the 980ti's were boosting to 1400mhz.  Using precisionx, lowest gpu offset it allows is -90, and it brought my gpu clock down to 1300mhz.  I was able to play 7 consecutive conquest games (with a break in between, of course) without crashing!  This is with all my cpu/ucnore/ram at highest stable OC setting too.  I'll continue to run it this way for a while to make sure everythings straightened out, but haven't gone this long without freezing up.
 
With that said, is it normal for the 980ti classifieds to boost to 1400mhz and not be able to sustain this gpu core speed?  Odd that I need to downclock it to keep them stable.

Had a good streak going, but game locked up again.
 
Going to try the -90 gpu clock offset WITH max voltage and power target using precision X.  Hoping it will help stabilize it.
 
Questioning the voltage control and power target though, because every time I adjust it up or down, still reports same voltage, is this normal?
 
Also one card is reading at 1.20V, the other 1.15V, maybe because of the difference in ASIC quality?

Use msi afterburner to underclock the card. Afterburner will allow more than a -90 underclock.
 
Adjusting power target doesn't make the card draw more power. Increasing the power target only allows the card to draw more power if needed.
 
1.212v is the max allowed voltage on the 980 ti classified stock vbios. Yes, different asic quality's can affect the voltage. Software readings can also show incorrect voltage readings.

At this point I suggest doing all of what I suggested in post #2.

All,
 
I picked up a multimeter and used probe it cables to check out actual voltages from the card directly.  Sure enough, GPU-Z and HWiNFO did not read the correct voltages.  PrecisionX was able to read voltages over 1.15-1.20, but still not as accurate as the multimeter.  I can confirm that the PrecisionX tool can bump your card up to 1.30V by turning on both overboost and overvoltage in settings.  I can also confirm that increasing power target percent above 100% does not get you any more.  See test results in attached photo.
 
Going to try and game with my 4.4Ghz overclock with stock 1400mhz boost clock, this time at 1.30V.  Hoping the extra voltage can keep it stable.  Perhaps, even go beyond 1400mhz.
 
 

Thanks for the follow up, Sajin.
 
So my OC at 4.4ghz with increased voltage on the gpu's also froze up.  Which leaves me to believe that it was my cpu OC and not the card that is the issue.  I am happy with settling with a 4.3ghz oc, with my RAM running at rated speed, and my uncore OC'd too, so long as this freezing business stops, heh.  To really prove cause was cpu OC, i can try and run the 4.3ghz oc with default voltage on the gpu's.  But it's so stable right now, thinking to leave the gpu's running at 1.3V 24/7.  If continue to stay stable, possibly even start OC'ing the gpu's.  Is 1.3V daily too much?  My temps don't break 60C on the gpu's.

Sajin, thanks for your help.  Saved me from RMA'ing my cards.
 
Your suggestion about using the bottom card worked.  however, it didn't improve temperatures even though the card running at lower temp was taking more load.
 
Gaming has been solid.  So solid that I started overclocking my 980ti's.  Got them up to 1455mhz on core, 3712mhz on memory.  Any higher, I start to experience tearing and driver errors.  Tried lower voltages, but not as stable as 1.3V.
 
One minor annoyance is that after booting up or restarting, the overclock in precisionx doesn't stick.  It drops the voltage to 1.29 instead and sometimes the memory clock offset on one of the cards will reset back to 0.  Sometimes it takes a couple of attempts and playing around with lowering the voltage and linking/unlinking cards is needed to get the memory offset to stick.  Weird.
 
Thanks again.