Is my card broken?

[Joeysaurr]

Been getting very regular kernel crashes recently. I've tried purging the drivers and installing fresh ones multiple times, I've also roller back to several different versions. Each time making sure the current drivers have been completely removed.

Sometimes I can go a few hours without any crashes but other times I'll get a black screen for up to a minute and when it comes back I'll have about twenty different notifications reporting kernel crashes.

I have an MSI GTX 970 4G and it's still under warranty so RMA'ing is an option but I've heard it can be quite a hassle, is there anything else I could try?

Other Specs:
i7 3770 @4.01 GHz
EVGA Supernova G2 850W
Gigabyte GA-Z77X-D3H

 

[Bossyfins]

If it just a driver crash, then it is drivers. Wait for NVIDIA to fix.

 

[bignastyid]

Could be many things. Posting the full system specs(make and model of the psu included) would be helpful.

 

[RaulKO]

Card is not broken since if it was the card will not work at all.

Most likelly the issue is that your card cannot keep stability with the stock OC that it has. Maybe this situation has been accentuated with the tipical warmer temperatures of summer.

I would recommend to try a small downclock of the core and see if the situation improves.

Of course you can always RMA the card but the downclock could work.

 

[Joeysaurr]

I doubt it's a true driver issue because I've been using drivers from way before this was an issue. I also doubt it has anything to do with clock speed because this is mostly happening on the desktop where I'm running at idle clocks.

 

[jason3871]

I doubt it's a true driver issue because I've been using drivers from way before this was an issue. I also doubt it has anything to do with clock speed because this is mostly happening on the desktop where I'm running at idle clocks.

Have you tried rolling back to previous drivers? Uninstall drivers using DDU- https://www.google.co.in/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CB0QFjAAahUKEwiDp_jbqeDGAhWVCY4KHX3xC5I&url=http%3A%2F%2Fwww.guru3d.com%2Ffiles-details%2Fdisplay-driver-uninstaller-download.html&ei=AP6nVYOnCpWTuAT94q-QCQ&usg=AFQjCNEJ9nAYe5bakSWLix-qMCQn0Fvqbw&sig2=3_kp9n_ulajMBWXRmdIFxA&bvm=bv.97949915,d.c2E Also ensure that your clock aren't running at 3D clocks when in supposed idle mode.

 

[Joeysaurr]

Yes, I've rolled back several official releases each time using DDU in safe mode to remove the drivers.
My idle speeds are 400-800 MHz (800 when gsync is enabled)

 

[jason3871]

Yes, I've rolled back several official releases each time using DDU in safe mode to remove the drivers.
My idle speeds are 400-800 MHz (800 when gsync is enabled)

Could you try increasing your GPU voltage. If you do that then in 2D clocks the voltage will increase as well. This usually happens when the voltage is too low.

 

[Joeysaurr]

I'll try increasing the voltage when I get home but my max power is already at 110% and it seems odd that my settings are suddenly an issue when I've had them since October last year and I've only been getting the crashes for two months
Is it overclocked by any chance?

 

[Joeysaurr]

Yes my clocks are +165 on the core, gpu boost takes that up to 1485MHz and +250 on the memory. I don't think the overclock is the issue though because if anything the crashes are less frequent under load.

 

[jason3871]

Yes my clocks are +165 on the core, gpu boost takes that up to 1485MHz and +250 on the memory. I don't think the overclock is the issue though because if anything the crashes are less frequent under load.

It wouldn't hurt if you went back to stock and checked if the crashes actually stopped.

 

[Joeysaurr]

No change with stock clocks.

 

[jason3871]

No change with stock clocks.

Did you try increasing voltage?

 

[Joeysaurr]

This was all done on desktop just browsing forums and YouTube.
+10mv: crashed once after ~20 minutes
+20mv: didn't crash after an hour
+30mv: crashed 7 times after ~40 minutes
+40mv: kernel immediately crashed 15 times.
+50mv: didn't crash after an hour.
+60mv: didn't crash after an hour.
+75mv: temp gone up by 5°C, crashed 3 times after ~50 minutes.
+87mv: crashed 3 times after exactly 37 minutes (sat and watched the clock that time)

Going to try +20,50 and 60 for longer next.

 

[jason3871]

So far on this forum I've come across numerous threads which have been having a similar problem that you have been having and all of them seem to have Nvidia GPU's which leads me to believe that it could very well be on a software level. Have you tried using DDU to uninstall and re-install much much older drivers? DDU- http://www.guru3d.com/files-details/display-driver-uninstaller-download.html

 

[Bossyfins]

Does it show a poppu saying driver crashed near the date and time?

 

[Joeysaurr]

Voltage doesn't seem to be making a difference.
I've tried as far back as September last year which is the earliest maxwell driver.

And yes bossy, it does.

 

[Bossyfins]

That is probably just drivers. NVIDIA recent drivers have been crashing, so before RMA card or anything like that, see if next driver fixes it. I get that too.

 

[delrocks22]

Download DDU or Driver fusion and do a clean uninstall of ur present drivers then download the previous version nvidia drivers and install them and see if the problem persists...

 

[Joeysaurr]

I've used DDU many many times. Installed dozens of different drivers and nothing has changed the fact that I get this crash up to 50 times in a 24 hours period. And mostly while idling.

 

[delrocks22]

I've used DDU many many times. Installed dozens of different drivers and nothing has changed the fact that I get this crash up to 50 times in a 24 hours period. And mostly while idling.

Try a clean reinstall of OS then and as soon as its done install the drivers which worked for u for the longest time period...and when nvidia launches next version, update that time....i dont think this is a hardware issue...so by formatting windows it should fix ur prob....
stuff u can try:
1. driver fusion to remove integrated and nvidia drivers completely
2. uninstall drivers in safe mode
3. install ccleaner and analyze and fix registry issues
4. scan for virus maybe?

If i was in ur place i would get rid of it by formatting everything and reinstalling windows lol