3 titan X 
3 Ek blocks w/ backplates
All 3 cards tested with air playing GTA V would run 1400/2000 fan @ 100%
Added water blocks and back plates checked thermal paste pattern before installing cards into loop.
one of the 3 water blocks did not make good flat contact redone with thermal paste and increased how tight i screwed the 4 bolts around the gpu, now makes contact even tho pattern is not 100% even at least is touching.
 
Installed all 3 gpu on asrock X79 Extreme11 mobo (dual plx + lsi SaS on board) Thermaltake 1350w psu (known good) IvyBridge-E 4960X @ 4.6ghz w 1.36v is stable @ 4.7. 32GB gskill 2400 10,12,12,31,240 T2. Gpu connected to 3 Korean refurbished Samsung monitors (Zevroid - 30Qu/z) 1600p via DVI-D with super shielded monoprice cables.
 
Now when i first installed them everything worked as expected. Except over clocking was worse under water than it was on air stock cooling gpu could still go 1350core and +200 mem (instead of the +500 granted this was only playing gta5 for short ish amount of time around 30min on each card.) with a little bump in vcore things got worse! Temps are great across all 3 gpu 32c worst case max temp. So ran them mostly at stock for a little over a month or so started playing ARK evolved and wanted a little extra lol! So back to OCing the gpu mind you only one gpu is used for ARK i settled for 1358core and +200 on the memory anything more would result in the ocasional gpu driver recovery or game crashing at first... then i started geting black screens non recoverable had to hold power to shut down. I first thought was maybe just that game or maybe just gpu3 was being used for that game for what ever reason (playing in windowed mode on center screen must have been on gpu 3). Untill i was playing wow on all 3 screens at stock gpu speed and bam all 3 screens go black can still hear on mumble (ventrilo/Teamspeak) would not recover had to press and hold button to power down... Now i know i have problem so i take all 3 gpu out and inspect under water blocks everything looks good even on the number 2 card who i was suspect of from start not properly touching with out extra force. So I reinstall GPU #1 the only card i used for some time on air before water blocks finally arrived. 
 
So on #1 gpu with water block stock speed i loaded up BF4 naval assult map the one with the sub inside the building? and after a min i get a black screen but only for a few sec then come back like nothing ever happened. So i keep playing it does it again after a few more min say 5min then just randomly after that. So i went in and applied my gpu oc that was for sure stable on air 1400core +400memory aka 1950 with a slight vcore bump +31mv game runs perfect for next couple hours until i got angry enough at the game to not play it anymore lol (Rage Field 4 FTW).
 
I am at a loss i feel like this gpu is acting something similar to what my 290x's were and need extra aux voltage for imc on gpu to be 100% stable unfortunately Titan X you can not change with EVGA Precision X. Now this is by no means my first use of water blocks or water cooling i have probably built over 50 complete water cooled systems over the last 10+ years and have never seen this type of behavior under water so i must be missing something i really wish i could see vrm temps on the nvidia gpu like i could on the 290x at least then i would know for certain they are getting proper contact from my water block. Everything looks good and the fact that running my over clock actually made my game more stable is just silly to me. Is this a known issue with BF4? Other games i play seem to be fine @ stock settings with 100% gpu load @ 1600p.
I know there is a Titan X owners thread that is 1200 pages!!!!! omg lol.  I seen some users talking about black screen issues maybe someone who has been following the thread can help me out i am not sure i can read that much all at once i my just lose my mind :P
 
Anyways if anyone has thoughts or ideas love to hear them.

Did you read? I stated BF4 did this only at stock clocks not overclocked.

formula383
3 titan X 
3 Ek blocks w/ backplates
All 3 cards tested with air playing GTA V would run 1400/2000 fan @ 100%
Added water blocks and back plates checked thermal paste pattern before installing cards into loop.
one of the 3 water blocks did not make good flat contact redone with thermal paste and increased how tight i screwed the 4 bolts around the gpu, now makes contact even tho pattern is not 100% even at least is touching.
 
Installed all 3 gpu on asrock X79 Extreme11 mobo (dual plx + lsi SaS on board) Thermaltake 1350w psu (known good) IvyBridge-E 4960X @ 4.6ghz w 1.36v is stable @ 4.7. 32GB gskill 2400 10,12,12,31,240 T2. Gpu connected to 3 Korean refurbished Samsung monitors (Zevroid - 30Qu/z) 1600p via DVI-D with super shielded monoprice cables.
 
Now when i first installed them everything worked as expected. Except over clocking was worse under water than it was on air stock cooling gpu could still go 1350core and +200 mem (instead of the +500 granted this was only playing gta5 for short ish amount of time around 30min on each card.) with a little bump in vcore things got worse! Temps are great across all 3 gpu 32c worst case max temp. So ran them mostly at stock for a little over a month or so started playing ARK evolved and wanted a little extra lol! So back to OCing the gpu mind you only one gpu is used for ARK i settled for 1358core and +200 on the memory anything more would result in the ocasional gpu driver recovery or game crashing at first... then i started geting black screens non recoverable had to hold power to shut down. I first thought was maybe just that game or maybe just gpu3 was being used for that game for what ever reason (playing in windowed mode on center screen must have been on gpu 3). Untill i was playing wow on all 3 screens at stock gpu speed and bam all 3 screens go black can still hear on mumble (ventrilo/Teamspeak) would not recover had to press and hold button to power down... Now i know i have problem so i take all 3 gpu out and inspect under water blocks everything looks good even on the number 2 card who i was suspect of from start not properly touching with out extra force. So I reinstall GPU #1 the only card i used for some time on air before water blocks finally arrived. 
 
So on #1 gpu with water block stock speed i loaded up BF4 naval assult map the one with the sub inside the building? and after a min i get a black screen but only for a few sec then come back like nothing ever happened. So i keep playing it does it again after a few more min say 5min then just randomly after that. So i went in and applied my gpu oc that was for sure stable on air 1400core +400memory aka 1950 with a slight vcore bump +31mv game runs perfect for next couple hours until i got angry enough at the game to not play it anymore lol (Rage Field 4 FTW).
 
I am at a loss i feel like this gpu is acting something similar to what my 290x's were and need extra aux voltage for imc on gpu to be 100% stable unfortunately Titan X you can not change with EVGA Precision X. Now this is by no means my first use of water blocks or water cooling i have probably built over 50 complete water cooled systems over the last 10+ years and have never seen this type of behavior under water so i must be missing something i really wish i could see vrm temps on the nvidia gpu like i could on the 290x at least then i would know for certain they are getting proper contact from my water block. Everything looks good and the fact that running my over clock actually made my game more stable is just silly to me. Is this a known issue with BF4? Other games i play seem to be fine @ stock settings with 100% gpu load @ 1600p.
I know there is a Titan X owners thread that is 1200 pages!!!!! omg lol.  I seen some users talking about black screen issues maybe someone who has been following the thread can help me out i am not sure i can read that much all at once i my just lose my mind :P
 
Anyways if anyone has thoughts or ideas love to hear them.

 

formula383
Did you read? I stated BF4 did this only at stock clocks not overclocked.

I did read your post. According to the OP "stock" means stock gpu clocks. You didn't mention that you ran your cpu & ram at stock as well.

I think it could be my PSU its a few years old now and according to the thermaltake psu calculator i am under power requirements for 3 cards that is with the 20% age on capacitors. That still would not explain the issue with one gpu installed but i'm going to replace it with a evga 1600w titanium anyways it has good reviews and idk maybe the titan X is just very finicky on clean power or lower voltage. This psu never had a strait 12v on the 12v rail, was always a bit low since it was new. Also i think i am going to move my raid card and 6 of the sas drives to another pc i just dont need to have that much extra going on inside this pc. Also I may add in a new dedicated 20amp line for my pc. Not like haven't popped enough 15amp fuses already lol. The pc it self only pulls 1400w from the wall max but you start adding fans and water pumps and monitors to the same circuit it really does not take much to pop a 15a fuse at all. And the new psu will be a bit better power efficiency so i should be pulling a bit less power total. Maybe I will look into adding a few caps inline too... I know when i play games that do not use that much power can really make the lights flicker when the gpu's go on and off very quickly.