Computer Crashes when gaming / rendering using SLI!

[LeviKosters]

First of all I would like to thank anyone that opened this.

I build my first PC a few days again, and I have noticed it to crash rather often. I am a 3D artist that has to do a lot of rendering in various 3D programs but when it gets to the case, my PC and / or software crashes.

Yesteday I was studying for an exam and my screen went black saying I had no input or cable connected. When I checked my case, both my GPUs were off and my system would not display. I took out one of both cards and it worked fine, even rendering went flawless but as soon as I put the other GPU back in shit would hit the fan quite literally.

All the Drivers are up-to-date, turning OC off does not help and my motherboard does support SLI. I have also noticed that the usage would never go up when rendering (utility of about 3%, no temperature rising and my fans never turn on), this goes for both gaming and rendering. These are the specifications:

- Intel Core i7-5930k 3.5GHz 6-Core Processor [OC at 4.4GHz]
- ASRock X99 Extreme4 ATX LGA2011-3 Motherboard
- Crucial 16GB (2*8) DDR4-2133 Memory
- [2x] NVIDIA GeForce GTX 970 4GB Twin Frozr V Video Card on SLI.
- EVGA 850W 80+ Gold Certified ATX Power Supply
- Crucial MX100 512GB 2.5" Solid State Drive
- Western Digital Caviar Blue 1TB 3.5" 7200RPM Internal Hard Drive
- Corsair H100i 77.0CFM Liquid GPU Cooler.
- BenQ RL2455HM 60Hz 24.0" Monitor
- NZXR S340 ATX Mid Tower Case
- Windows 8.1 Professional

 

[emdea22]

You need to test those two cards individually.
Use FURMARK for 20mins and Unigine Valley for about 30 min for GPU testing
Use Shadow of mordor or battlefield to max out vram for video memory testing
First test individually and then in SLI.

In case you get no problems with the above tests you might want to try downloading Dota2 and just watch some replays for 1 hour. I've found dota2 to quickly show an unstable card even tho it may pass other tests.

If everything passes using a single card then its either the heat, a software(even bios) conflict or bad PSU

 

[LeviKosters]

You need to test those two cards individually.
Use FURMARK for 20mins and Unigine Valley for about 30 min for GPU testing
Use Shadow of mordor or battlefield to max out vram for video memory testing
First test individually and then in SLI.

In case you get no problems with the above tests you might want to try downloading Dota2 and just watch some replays for 1 hour. I've found dota2 to quickly show an unstable card even tho it may pass other tests.

If everything passes using a single card then its either the heat, a software(even bios) conflict or bad PSU

Test results came out fine. On a single GPU I had benchmarks up to 80FPS in FurMARK and 110 in Uningine Valley and on SLI I got results going up to 140FPS in FurMARK and up to 200FPS in Uningine Valley. I've played Battlefield before on SLI and singlehanded and that went fine as well, tested it on SLI just to be sure and hit a solid 170FPS.

What do you mean with heat? What could I do about Software and / or how can I improve my PSU?

 

[emdea22]

two cards in SLI heat your case alot. That heat is not only affecting your GPU but also motherboard cpu etc. So due to excessive heat from the GPUs your PSU (only if mounted on top) and cpu vrm area get overheated resulting in system instability. Also its possible to overheat motherboard chipset which shouldn;t exceed 60C at most. The way you solve this is you get extra cooling for the case especially near the vrm area of the cpu.

Theres a reason why SLI configs aren't ideal and the above is just one of them. Software conflicts can mean a variety of things:
-driver corruption (try a clean reinstall )
-3rd party software conflicting with driver (MSI afterburner and another OC tool; any other software conflicting with drivers)
-SLI unsupported properly by your rendering program (may be limited to one or more GPUs)
-in case of IGP -> conflicting drivers (disable IGP from bios and purge igp drivers)
-different bios on cards (should contact the manufacturer on how to solve)

PSU can either overheat and shutdown or it has a loose connection. Keep in mind these new cards from both nvidia and AMD have weird spikes that aren't that well supported by the PSUs. There was an article on toms about that not too long ago. Make sure you check the ends of the cable since your PSU is modular and make sure they make contact properly.


Bottom line get your case well ventilated on any SLI setup - it might spare you lots of issues.

 

[LeviKosters]

two cards in SLI heat your case alot. That heat is not only affecting your GPU but also motherboard cpu etc. So due to excessive heat from the GPUs your PSU (only if mounted on top) and cpu vrm area get overheated resulting in system instability. Also its possible to overheat motherboard chipset which shouldn;t exceed 60C at most. The way you solve this is you get extra cooling for the case especially near the vrm area of the cpu.

Theres a reason why SLI configs aren't ideal and the above is just one of them. Software conflicts can mean a variety of things:
-driver corruption (try a clean reinstall )
-3rd party software conflicting with driver (MSI afterburner and another OC tool; any other software conflicting with drivers)
-SLI unsupported properly by your rendering program (may be limited to one or more GPUs)
-in case of IGP -> conflicting drivers (disable IGP from bios and purge igp drivers)
-different bios on cards (should contact the manufacturer on how to solve)

PSU can either overheat and shutdown or it has a loose connection. Keep in mind these new cards from both nvidia and AMD have weird spikes that aren't that well supported by the PSUs. There was an article on toms about that not too long ago. Make sure you check the ends of the cable since your PSU is modular and make sure they make contact properly.


Bottom line get your case well ventilated on any SLI setup - it might spare you lots of issues.

Disabled OC and did some manual updates on the drivers. Let Windows check if the Drivers were up-to-date using the Device Manager and did the regular Windows Updates. Turns out my Windows Update had turned itself off somehow and I was missing out on some important driver updates (about 64).

TL;DR Don't be a rookie people.