So I finally got my new rig up and running only to find out that my video cards are causing BSODs! At first they were only crashing when I overclocked them with EVGA PrecisionX 16 to 110% power and 100Mhz GPU Clock Offset and they were happening in 10-30min intervals on Witcher 3 on 4k almost max settings. But now I just got a BSOD while only setting the Power Target to 110%, which never happened before, so I decided its time to ask the Pros for help!
Also, I have to run Witcher 3 in Fullscreen borderless mode, otherwhise it'll blink green continuously (is this micro-stutter or flickering?)
 
This is my latest WhoCrashed Crash Dump details:
Crash dump file:        C:\Windows\Minidump\082815-18080-01.dmp
Date/time:              28/08/2015 04:12:36 GMT
Uptime:                 02:30:54
Machine:                THIAGO-PC
Bug check name:         
Bug check code:         0x50
Bug check parm 1:       0xFFFFF88025223000
Bug check parm 2:       0x1
Bug check parm 3:       0xFFFFF880065C652F
Bug check parm 4:       0x0
Probably caused by:     
Driver description:     
Driver product:         
Driver company:         
OS build:               Built by: 7601.18939.amd64fre.win7sp1_gdr.150722-0600
Architecture:           x64 (64 bit)
CPU count:              12
Page size:              4096

Bug check description:
This indicates that invalid system memory has been referenced.

Comments:
This appears to be a typical software driver bug and is not likely to be caused by a hardware problem.
A third party driver was identified as the probable root cause of this system error. It is suggested you look for an update for the following driver: nvlddmkm.sys (NVIDIA Windows Kernel Mode Driver, Version 355.60 , NVIDIA Corporation). 
______________________________________________________
Some older one:

 
Crash dump file:        C:\Windows\Minidump\082315-18720-01.dmp
Date/time:              23/08/2015 17:13:09 GMT
Uptime:                 00:44:22
Machine:                THIAGO-PC
Bug check name:         
Bug check code:         0x50
Bug check parm 1:       0xFFFFF880241B6000
Bug check parm 2:       0x1
Bug check parm 3:       0xFFFFF8800660B52F
Bug check parm 4:       0x0
Probably caused by:     
Driver description:     
Driver product:         
Driver company:         
OS build:               Built by: 7601.18933.amd64fre.win7sp1_gdr.150715-0600
Architecture:           x64 (64 bit)
CPU count:              12
Page size:              4096

Bug check description:
This indicates that invalid system memory has been referenced.

Comments:
This appears to be a typical software driver bug and is not likely to be caused by a hardware problem.
A third party driver was identified as the probable root cause of this system error. It is suggested you look for an update for the following driver: nvlddmkm.sys (NVIDIA Windows Kernel Mode Driver, Version 355.60 , NVIDIA Corporation).


__________________________________
 
I can link more if needed, but they all seems the same to me, always pointing to nvlddmkm.sys.
I already did an 8pass MemTestx86 overnight and not a single error was found.
Also tried using DDu reseting to the *latest* nvidia drivers, with no luck. Still havent given a try to older drivers.
Also tried reversing the sli bridge or going for a different connector (the GTX 980 Ti seems to have 2 options for these on a 2 sli configuration). Didnt try all combinations.
Didnt try another sli bridge, dont have one in hand right now.
Nvidia control panel SLI is set to "Maximise 3D settings" (sorry if this is the wrong name, my OS is not in english)
Already got  power management in Nvidia Control Panel to "Maximum Performance" and added Witcher 3 to "Manage 3D Configuration", also setting its power management to Maximum Performance, with SLI Render Mode set to "Recommended by NVIDIA (SLI)". (Although I havent ran any tests after it)
Did one Chkdsk /R /X for each HD/SSD and they all prompted that Windows didnt find any problems.
 
This is my first time with a SLI configuration, so I dont really know alot about it.
 
Its only my second week with this rig.
Since my HD/SSDs are older, I decided to do a fresh Windows 7 x64 Professional install (formatted Games SSD and Data HDs, but not the Main SSD), then going for the drivers that came with the hardware, following to Windows Update and then deleting Windows.old folder while keeping some files.
 
Havent updated BIOs so far.
Got Corsair Link and Kaspersky installed (though it could come in handy).
Havent noticed any overheating.
Power settings set to High Performance.
PhysX configuration is set to "Automatic".
 
Twice I tried to follow a fix with this nvlddmkm.sys file and twice I got a new BSOD and had to do a system restore to get my OS to run again.
 
With that being said, here are my system specs:
1x GTX 980 Ti Classified - new
1x GTX 980 Ti Hybrid - new - (Yeah, got two different ones but I kinda didnt have a choice here and EVGA rep told me it was fine) - ( stock fan which is set to exhaust, installed in the back)
The Hybrid is installed on the first PCI-e slot while the Classified is installed in the third one.
My TV and main monitor are connected to the Hybrid card, while the Classified have a second monitor plugged in (new DisplayPort cable coming soon so it will all be installed on the Hybrid).
Should I just connect all of them to the Classified instead? Does it make any difference? Ive read that they should be connected to the coolest GPU, which is why I got them to the Hybrid.

Motherboard: EVGA X99 FTW - new
CPU: Intel Core i7 5930k (no overclock) - new
RAM: 8x Corsair 4gb DDR4-2133 CMK16GX4M4A2666C15 (no overclock) (half of those appear with a different clock settings during BIOs, dont know why) - new
Radiator: Corsair H110i Gt stock fans - new (exhaust, top)
Power Supply: Corsair ax1200i (1200w, 80 Platinum)- about 1.5 years old but used for about 6 months and it stood stored for the rest of the year
UPS: CyberPower OR1500PFCLCD PFC Sinewave 1500VA 1050w (yeah, its lower then my Power Supply, but even with my TV plugged in it never went past 800w)
Case: Corsair 780T - new
Main SSD: Samsung 840 PRO Series (256GB) - 2 years old
Game SSD: Samsung 830 Series (256GB) - 3 years old
Data HD1: WDC WD1001FALLS-00E8B0 1TB - 3 years old
Data HD2: ST300DM001-1ER166 - new
Fans: 2x - 140mm NT-FZ-140-U NZXT (front, intake)
         1x - 120mm NT-FZ-120-U NZXT (bottom, intake)
 
TV: Samsung 55hu8500 (4k) - 1 year old
Monitors: 2x Benq xl2420t 120hz - 2, 3 years old
 
Using a bunch of peripherals.
 
 
 
So, what can I do to stop this crashes? Actually, what is causing them?
 
Extra question: Should I really be getting only ~40fps on Witcher 3 4k with most of the settings set to the best possible? With hairworks this goes to about 30fps.
 
Let me know if Im missing any worthy information!
 
Thanks!
 
 

They are based off the same gpu, it really shouldnt be a problem?

Now that you mentioned, not really. Both are set to their stock, Hybrid to 1139Mhz and Classified to 1189Mhz. Ill try to match those and see what I can get. Should I mess up with the voltages aswell?

Try Afterburner or precision to see if you can get them to link up.

thrmotta
 
RAM: 8x Corsair 4gb DDR4-2133 CMK16GX4M4A2666C15 (no overclock) (half of those appear with a different clock settings during BIOs, dont know why) - new

That is your problem right there. You need to setup your ram's frequency, timings & voltage to match the jedec stickers on the sides of your ram modules.
 

 
Updating the bios should correct the problem. If it doesn't you will need to input the settings manually.

I finally got some spare time to try those fixes, so Ill be updating bios first, adjusting the memorys if needed and see how it goes.
 
Ive never flashed a BIOS to a videocard before, so Ill be reading a few tutorials first before trying anything. If it comes to this, should I flash the Hybrid to Classified or vice-versa? What will be the difference? In my naivety Id flash Hybrid to Classified once the Hybrid should be able to handle overclock better while still mantaining some good (if not great) lifespan.

And just to be sure, flashing bios in this case would only match the clock of both cards, which is the same that can be achieved with PrecisionX, but would save me sometime at the cost of a "permanent" overclock. Is that it?

Thanks!!

JerkMan

thrmotta
Now that you mentioned, not really. Both are set to their stock, Hybrid to 1139Mhz and Classified to 1189Mhz. Ill try to match those and see what I can get. Should I mess up with the voltages aswell?

Depends. You're safe adding voltage via software, but it'll add heat. If I were you, I'd flash the Classified BIOS to the HYBRID, it should handle it and match your clocks, then go from there.

I would Not flash the bios from a card that has completely different VRM setup to another card.. I would use maxwell bios Tweaker (available on overclock.net) to adjust a 980ti hybrid bios to match the classified clocks.

The new Bios along with setting both cards to the same clock seems to do the trick! Ive played The Witcher 3 for about 3 hours straight with no crash/artifacts and FPS around 44.

Thank you all for the help!