Hello everyone,
    I just built a new PC recently and encountered a annoying problem. The major parts are:
CPU: i7-6700k (no overclock)
MB: ASUS ROG MVIII ranger
RAM: G.Skill, 2*8G DDR4-3000
Video card: EVGA GTX 970 SSC ACX2.0+
Cooler: Corsair H90
PSU: Corsair RM850 850W
SSD: samsung 850 evo  HDD: Seagate 1.5T 7200rpm
OS: win7 64bit
   All the parts were bought from newegg except the HDD which is from my 5 years old PC.
   I found this problem when I begin to use it and it is very reproducible. When I start gaming or start 3d mark 11 test, my computer will shut down after ~3-5mins. There are no error codes or warning about overheating parts. The power just shut off and I can re-power on my computer immediately after that.
   I tried to diagnose the problem. Firstly, I ran some CPU stress test using Prime95 and see if it will trigger the problem. It doesn't. I ran Prime 95 for 20mins and the CPU heated up to 70C without any problem. So I feel CPU should not be the source.
  Then it comes to two of my major "suspect", video card or PSU. I borrowed another computer from my friends. The major parts are:
CPU: i7-4970k (no overclock)
MB: gigabyte z97x-gaming
RAM: G. Skill, 2*8GB DDR3
Video card: EVGA GTX 970 FTW w/ACX2.0
PSU: EVGA supernova 650W? (spec was covered so I cannot see.)
SSD: samsung 850 evo
OS: win7 64bit
   I swap the 2 video card from the 2 rig and ran gaming or 3d mark 11 to see the stability. Both are fine and pass the 3D mark 11 2 times without any problem. When I put my SSC card into my friends rig, the max temp went to 75C under default fan profile. His FTW card went to 67C in my rig. At least none of them trigger any sudden power off of the computer, so I feel the video card should not be the cause of my problem.
   So finally I am focusing on my PSU. Initially I don't think PSU should be a problem since it is a corsair and it is a 850W. It should be much more than enough for my system. Also it can power up the FTW without any problem. But Then I notice a maximum power drawing difference of the 2 cards. The SSC cards will draw 201W and the FTW draw 170W. At the same time, I post a thread on toms hardware and they said it will be more likely due to a defect PSU. So I feel it might be true so I took my 5 year old antec 650W PSU from my old rig and put in on. The problem get a little bit better. It did pass the whole 3d mark extreme test once. But when I ran the test multiple times, the problem will happen again but at least it can pass several scenes in 3d mark 11. I need to mention that initially it cannot even pass the first scene when I use RM850. 
   So I feel I should start to blame the PSU. It seems that it cannot hold a constant power and it may trigger some kind of protection when the video card is on load. So it will power off everything. But I also feel frustrated on the 2 video cards I have right now. The FTW w/ACX2.0 should be better than my SSC w/ACX2.0+. But it will consume less power and it ran cooler than my SSC. I thought the ACX2.0+ should at least be better in cooling. But according to my test, it ran ~10C higher than the FTW ACX2.0. Also when I put the FTW into my system everything works fine, no power off problem so far. It suggests that the PSU can hold the FTW while it cannot hold my SSC card which may draw ~30W more power max. I am not sure if it is a normal results or it also suggest that my SSC card also has some problem?
   
    I hope someone from EVGA can help me on this since I have spend a lot of time on that problem and I feel really frustrated.
  

All the temp was read out from the hwmonitor and I did not found any overheating issue. I did not really pay too much attention to the 12Volt rail.  But I feel it may trigger some kind of protection so it power off my system.
 
I am currently ran it without side panel so I can monitor the inside more easily. But I can add a side case fan and that is not a problem.
 
I still feel not quite right since the FTW w/ACX2.0 ran much cooler than my SSC w/ACX2.0+ with almost same setup. Also for same model (for example FTW ACX2.0 and FTW ACX2.0+), the card with ACX2.0+ will draw more power than ACX2.0 while I would assume the performance remain the same?? I just feel frustrated why the PSU can hold the FTW(170W) while it cannot hold my SSC w/ACX2.0+ (201W).
 
I have already RMA the PSU with newegg and will seek for another PSU to see if it can solve the problem.

Gotta say, sounds eerily similar to an issue I had with a Pc Power and Cooling PSU years ago. It is very possible a capacitor has failed in the psu. 
 
If that is indeed the case, it is an easy fix. But is is also possible that the wiring in the wall socket is faulty...not probable mind you, but possible. 

I feel completely losted. I bought an EVGA supernova 850 G2 PSU and return my corsair RM850. I feel shocked that the problem still exists. My computer still shut down when I ran 3d mark 11. So I took all my parts out of the case and test it outside. The situation get a little bit better. It may survive some scenes or even the whole test if I am lucky. Also I found if I use a custom fan profile to turn on the fan of my SSC card. It will shut down every time when I ran 3d mark11. The only time it survive the whole test, it was under default profile (without use precision X app) and the max GPU temp was read as 78C based on HWmonitor. But when I ran it again it failed and shut down my system.  I cannot switch out to see the temp when the 3d mark 11 is running so I don't know what the temp it is when it will shut down but I can feel from the card it is pretty warm.
It looks like I should not blame the PSU since I have tried 3 PSUs but none of them can survive. I feel realllly frustrated for this. Considering the FTW card can solve my problem I will have to return my SSC and find another one. But I still feel worried based on my previous thread. The SSC ran without problem in my friends rig. I am start to suspect that the problem may be my MB.

Removing the MB from your case would eliminate possible short to ground with the install, including back-plate ground finger finding its way into on of the ports.
 
MB mounting screws should be tightened lightly - too heavy of a hand and you can crush the MB - as it is several layers it could be damaged if excessively torquing those screws.
 
Recap of diagnostics
CPU - TIM and cooler were tested to be OK with Prime.  - 70C temps seen on CPU - OK
 
Your GPU works in your friends system. (usually means graphics card is OK)
 
Your MB / system works fine with your friends GPU. (power draw through PCIe port potential issue)
 
201W = 16.75 amps-your card
170W =  14.167 amps - friends card
 
New PSU and its included wires did not fix the issue.
 
Your MB does not have an auxiliary PCIe power connector.
         - So GPU pulls all its power from cables to PSU and PCIe bus.
 
You have confirmed all power connections to GPU and MB
 
You can fully seat GPU in PCIe slot - you visually inspected PCIe slot for damage and debris
 
PC runs fine under low stress uses - only fails when gaming or 3d mark 11 test - both high load on GPU
 
 
Test 1 - Reset your BIOS to default - this is an ASUS basic recovery method
Are you running the latest BIOS version?
 
Make sure you disable the on-board graphics in BIOS and turn off any features you do not use.
 
Set Graphics in BIOS to PCIe.
 ---  Run 3d mark 11 test software -- check for crash--
 
Test 2
Normally the GPU is placed into the slot closest to the CPU. (PCIe slot 1)
 
- Try moving your GTX 970 SSC to a different PCIe slot. 
 
If it work in a different slot it would prove that the MB primary PCIe slot has an issue that only surfaces under higher wattage loads.
 
Last ditch effort: remove and reload GPU drivers - only load video drives - custom install - test
 
If it still fails it would indicate MB issue under high wattage / amp loads   ---Contact Asus support

Thank you very much for your suggestions. I feel I am getting closer to the true cause of the problem.
I have remove and reload the drivers multiple times with different versions previously. I also tried switch the RAM slot to exclude RAM slot problems. Nothing works. I did not mention in my thread.
I update the MB bios to the lastest version today and do lots of 3d mark tests. I have some interesting observations. My computer looks more stable under default GPU fan profile (no precision X app) and quiet GPU fan profile (precision X app loaded) even it means higher temp. It can pass 3d mark 11 test today better than before under default GPU fan profile with no shutting down. But it is not stable under some more aggressive fan profile. When I ran the 3d mart 11, the fan speed may goes higher than 2000rpm. Then my computer may shut down after some time and it cannot pass 3d mark 11 test.
Then I switch the PCI-E slot and visually inspect the major slot for damage. But based on my view, I cannot see obvious damage on the slot, also on the "gold finger" of the card. Finally I was very shocked to find that after I put my card into slot 2 (3 slots total), the problem seems to be solved. I ran 2 consecutive 3d mark 11 test under aggressive fan profile and there is no shut down at all. I don't have time to ran some game test to see long term stability but I feel it may give me some suggestions that the problem may be due to the major slots of the MB?
I still don't know the detailed cause for my problem. But I feel if I want to truly solve my problem, I may need to replace the MB instead of the video card?
 

liyihaha
Thank you very much for your suggestions. I feel I am getting closer to the true cause of the problem.
I have remove and reload the drivers multiple times with different versions previously. I also tried switch the RAM slot to exclude RAM slot problems. Nothing works. I did not mention in my thread.
I update the MB bios to the lastest version today and do lots of 3d mark tests. I have some interesting observations. My computer looks more stable under default GPU fan profile (no precision X app) and quiet GPU fan profile (precision X app loaded) even it means higher temp. It can pass 3d mark 11 test today better than before under default GPU fan profile with no shutting down. But it is not stable under some more aggressive fan profile. When I ran the 3d mart 11, the fan speed may goes higher than 2000rpm. Then my computer may shut down after some time and it cannot pass 3d mark 11 test.
Then I switch the PCI-E slot and visually inspect the major slot for damage. But based on my view, I cannot see obvious damage on the slot, also on the "gold finger" of the card. Finally I was very shocked to find that after I put my card into slot 2 (3 slots total), the problem seems to be solved. I ran 2 consecutive 3d mark 11 test under aggressive fan profile and there is no shut down at all. I don't have time to ran some game test to see long term stability but I feel it may give me some suggestions that the problem may be due to the major slots of the MB?
I still don't know the detailed cause for my problem. But I feel if I want to truly solve my problem, I may need to replace the MB instead of the video card?
 

Yes the MB is the issue based on your information.
 
Glad to help you narrow it down to the PCIe slot 1.
 
The higher watt load of your card along with the fan issue were key to solving this mystery.
 
If it was just the GPU card, the issue would have continued in the other slots along with in your friends system.  The fan running at speed - custom profile - was an additional load (watts)
 
The test CPU with Intel Processor Diagnostic tool - suggestion from bsmegreg - can't hurt to try.