980 Ti Classy SLI Issues

[twisted46]

Hey all,

Let me start by saying I already searched and can't seem to find my specific problem, but maybe I am just bad a google.

Anyways my issues is that I just added a second 980 Ti Classy into my custom loop PC and am having a rather disappointing Issue. Computer boots fine, runs fine, does everything fine.. Except run any graphically demanding program. To expand on that, any benchmark i try to run causes the PC to restart almost immediately after loading in. There is no blue screen or frozen video, just instantly powers off and reboots.

PC Specs:

ASrock X99 Extreme 4
5820K @ 4.4 GHz
EVGA 980 Ti Classy x2 in SLI
4x4 Gb of Crucial 2400 DDR4
Corsair RM850 (Gold)

I am using EK water blocks on all components (full cover GPUs) the Mobo is not WC.

I am now learning that the RM850 is kind of a crap PSU (it was always fine with card and still is if SLI is disabled.

Things I have checked:

Temps, Everything is looking great just as it should with a custom loop, no spiking that may trigger a power off.

SLI settings, I made sure drivers were good and enabled SLI in Nvidia CP.

Overclocks, I took the clock off my CPU and no change, again it runs just peachy without SLI. No OC on the GPUs since the Classy boosts pretty high on its own.

Power draw, I dialed both the cards back to 80% to see if maybe the PSU was being maxed out, but I feel like a 850w PSU should be okay without crazy overvolting.


Let me know what you guys think.

Thanks,
T

 

[The_Staplergun]

Try using the new card by itself in the primary slot without the other card in and stressing it. If the system does the same thing it's the card.

 

[OneFIshSunfish]

I had a similar issue when I attempted to swamp around my SLI configuration (2 970s). DDU and reinstalling drivers fixed the issue. Could also be an issue with the bridge. Make sure It is seated properly.

 

[10tacle]

Run CPUID hardware monitor when under load like gaming and see if your PSU +12v rail is fluctuating beyond 5%. This is exactly how I discovered my Corsair TX750 PSU was dying on my 970 SLI rig when my PC was crashing during games (CPUID showed it dipping down to 10.7v).

 

[twisted46]

Quick Update and more info.

I was able to run the GPUZ stress test that puts both GPUs on max load and it ran fine for 15 mins so I am not thinking it is power draw now.

Also I am running at native 4K if that matters.
Running windows 10 Pro on a Samsung 480 SSD and all programs from a 3TB HDD cached to a 240 SSD.

The New Classy is the primary card in SLI and works great solo.

The problem with swapping GPUs is that They are already under water. I quickly did a bench test just to verify windows saw both cards and then rebuilt the loop.

For clarification the instant restarts have come from:

Firestrike as soon as the "Demo" loads. (same with Skydiver)

Heaven about 2 seconds after it starts

And recently tested Tomb Raider and it failed on the the start screen (very odd)

I will try all suggestions I can and report.

 

[twisted46]

So I tried DDU and no luck, I also hooked up the Molex aux PCIe power to my MOBO and no change.
Swapped out SLI bridges with no luck.

I do not see any Voltage issues.


Edit:

I am going to try after work:

Using Power cable that came with the psu (using a random one atm).

Hooking up a second psu to power one card.

Setting the card clocks to something crazy like 1000 Mhz and not letting them boost.

Bios flash on the board and both cards.

 

[twisted46]

Well this has gone totally off the rails. In an attempt to fix SLI issues I blew one of the cards because I was not thinking right at all, I am very disappointed in myself.

The worst part is that with the rig completely back apart I am not any closer to a solution for the original problem.

No idea if it is hardware or software or even which component was causing the problem. The blown gpu was a result of a factor i should never have tried.

 

[OneFIshSunfish]

what did you do?

 

[twisted46]

Kept trying to stress test and the secondary GPU popped with a nice plume of smoke about 10 seconds into heaven. I am wondering if MOBO power was the problem? I tried my Corsair CX650M in the rig too and it didn't change a thing. I should have eliminated all possibilities first... Now I have a dead card. Got to spend last night draining my loop and taking everything apart. Luckily the rig is back up minus the second GPU (obviously).

I tested the GPU in my server and it is for sure dead.

I really want to know what the issue was but I am very gun shy to give this another try.
May just cut my losses and sell my remaining card and get a 1080Ti when they drop.