GTX 970 Issue

Apr 13, 2018
10
0
10
So a few weeks ago my games started to crash, and by crash I mean my monitors would turn black and my PC would reset itself or I would have to force reset it. After i confirmed that it was not a heat issue or power issue, I deleted my drivers using DDU and re-installed them and all was well. A few days later the issue turned up again. These are the other fixes I have tried so far:

Connect graphics card to a different slot on my mother board.
Use only 1 monitor at a time.
Run stress tests for like an hour, and no crash.
Lower monitor rate to 60hz rather then 144hz.

Specs:
Intel I7-4790K @ 4.00Ghz
16Gb Ram
Windows 10
Gtx 970

Also good to note that i have a total of 3 monitors, one of which is connected directly to my motherboard and makes use of Intel(R) HD graphics 4600. If i set this as my main display and play games on it, they won't crash so it's definitely a nvidia issue.

Another thing I would like to point out is that my event logger does not always post the same issue, but they it's mostly a combination of either nvlddmkm 13 or DistibutedCOM 10016.

As you can see i have spent some time on this, so if anyone can help me in any way I would greatly appreciate it

Ps: This is my first time posting something on Tom's hardware so if i forgot something, my apologises xD
 

JoeMomma

Distinguished
Nov 17, 2010
860
1
19,360
I have a similar system (Z97 + i5-4690K + 16GB RAM) and had a similar problem when I had a GTX 970.
It turned out that the 970 was running out of the 3.5GB VRAM.
Due to a buggy game that would not clear the memory after it used it.

I could see a similar problem happening with 3 monitors. More monitors use more VRAM.
While getting a good PSU is always a good move, try using a single monitor as a test to see if VRAM is the problem.
You may want to get a video card with more memory. When I got a GTX 1070 with 8GB VRAM, the crashes stopped happening.

Your mileage may vary.
 
Apr 13, 2018
10
0
10


So I installed GPU shark, and i confirmed that it is not the issue, but however it does say GPU voltage limit reached, and GPU overvoltage limit reached, is this normal or?
 
Apr 13, 2018
10
0
10


I don't have the parts to do that unfortunately
 

JoeMomma

Distinguished
Nov 17, 2010
860
1
19,360

Hmmm . . .
IDK but maybe you can try to make and underclocking profile using MSI Afterburner?
I never had much luck with overclocking GPU's. It would always overheat and go back to normal.
At best you only gain a few percent better by overclocking a GPU.
Underclocking will only lose a few percent, but that's better than being unstable.

Try going into settings and check 'Unlock Voltage Control' and 'Unlock Voltage Monitoring.'
Make a Profile with minimum Core Voltage and minimum Power Limit.
Then set that Profile to be used in 3D situations.

 
Apr 13, 2018
10
0
10


Ok so to start things off i think you should know I never used MSI burner before, I checked 'Unlock Voltage Control' and 'Unlock Voltage Monitoring.' like you told me. My core voltage is at +0, and it can't go any lower then that, while my Power Limit is at a 100, if i try to lower it my temp limit also goes down, and same if i try to increase it. How high/low should i put it. To scared to just do things on my own tbh xD, don't want to fry my card

 

JoeMomma

Distinguished
Nov 17, 2010
860
1
19,360

As I said, I am no expert at GPU O/C either.
But I do know that by lowering the power limit you are underclocking and there is no danger.

The key with Overclocking or Underclocking anything is
patience, writing things down and making small adjustments until you get it right.

Usually it gets to the point in my overclock where it works, but might crash a few times, then I back off one notch.
My CPU overclocked to 4.6 GHz. It crashed once and I put it at 4.5 GHz. It never crashed again.

Try moving the power limit to 90%, then 85%, 80%, 75%, etc. A little lower each time.
When you get too low a voltage the card will start to crash because the chips don't have enough power.
It's crashing now at 100%, so you have to find the lowest point where it crashes and settle for a stable voltage in the middle.
Save separate Profiles for each setting. But don't apply those Profiles yet, just save them.
Always leave Profile #1 unchanged as your 2D setting or fall back to stock setting.

Then in Afterburner settings go to profiles and assign the 3D profile to Profile #2, #3, #4, etc.
Stress Test the card by running a bunch of things at the same time.
MSI Kombuster + Photoshop + YouTube + a game + a CPU stress test. I use OCCT.
You are trying to make it crash, by making it go faster than any sane person would do.
If it works for a crazy person like me, it will work for you. :pt1cable:

Once you get that working, you may also want to change your 2D Profile setting to one that you altered.
Imagine you could have 2D profile as 90% power limit and 3D profile as 85% power limit.
Usually when overclocking you go the other way 105% up to 110%.

It's almost impossible to destroy a card with one overclocking attempt.
Unless you are dumb and slide everything to the max on your first try.
It's not like they blowup, they usually lock-up and fail to boot or or it crashes all the time.
Then the GPU BIOS has to be reset to normal. I have never had to do that.
(Check your manual. Some cards have 2 BIOS. One to mess with and one as back up with an easy switch. Some have to be reset by shorting a jumper.)
Usually they only burn out or just wear out faster if you leave it at higher voltage than you should for a long time.

You are reducing voltage. You are underclocking.
It will wear out slower. Don't worry. You are fixing it. Not risking it.
 
Apr 13, 2018
10
0
10


Ok so after a while of testing, I sort of made progress. I decided to go down in factors of 5% and to use fortnite as a way to test my stability. I was not gonna try to stress test it as all stress tests always ran smoothly. After only just 1 downgrade (so here it was at 95% power limit) my game ran fine.

I thought that we had finally found the solution and my hopes were up. After an hour I moved onto portal 2, but unfortunately it crashed almost instantly. I kept on downgrading until i finally reached 70%. At this point i was fed up as the process is very tedious and annoying. ( Having to reset pc, wait for it to boot, and then wait for game to crash). At 70% the game did not crash for like 5 minutes so i decided to start playing, but after 5 minutes of shooting portals it crashed yet again.

Here my temper got the best of me and i put it to the lowest setting possible, which for msi afterburner is 62%. The game ran perfectly, I played for like an hour, maybe a bit more. I finally thought we had found a solution but ofc it was short-lived cause it ended up crashing.

It's currently 2 in the morning here so I'm going to call it a night but tomorrow evening I am going to try to do the same thing with my core clock instead, see if that has anything to do with it. My theory is that the gpu that i have is overclocked from the manufacturer, Gigabyte. Maybe if i down clock it to the original values that the 970 should have, I will get my fix.

This issue is really a though one and i don't get how it is happening because i have had my gpu for a while now ( I think over 2 years) so if this was a power issue wouldn't this have happened at the start as well?

 
Apr 13, 2018
10
0
10


Second Update:

I lowered both the core clock and memory clock by -40, but it crashed after an hour again. Should I try lowering it even further tomorrow or is there perhaps another thing I should try. My dilemma is, if i lower even further I might not get enough fps to keep my 144hz monitor up to speed.

Thanks in advanced
 
Apr 13, 2018
10
0
10


Third Update, I know it has been a while but have not really had to time to play games recently. I lowered it down all the way to -200 Mhz in both core clock and memory clock and it still crashes, sometimes instantly while other times it takes a while. I'm giving up hope here tbh, and help would be appreciated. Thanks in advanced.