GPU crashing ,Not heat related.

Salma_75

Distinguished
Jan 2, 2013
30
0
18,530
Hello :)

TL;DR : GPU drivers crashing on simple games , passed Furmark test but not Heaven Benchmark.


I have a problem with a PC that i purchased new 2 years ago.

Recently, Graphic card drivers started crashing while still loading games. Browsing the net & watching videos causes no problems.
It only crashes when trying to launch video games.
Things like Witcher 2 & minecraft. World of warcraft "a not so demanding game" would crash even if video settings were set to the lowest.

The error message is:
"display driver nvidia kernel mode driver stopped responding and has successfully recovered"

Card is out of Warranty & no one in my area got enough expertise to fix such problems.

*I Experience no Artefacts , white lines , black screens , blue screens or anything like that.
*Nothing was OC'd.
*No changes to software prior to the crash.


Things I tried so far:


-Monitoring Card's temps with HWMONITOR showed card crashing at 43c.
-Tested card for 15min with Furmark , no crashes , Temp stabilized at 83c.
-Attempting to test with Heaven benchmark crashes the program 5 seconds later.
-Tested CPU for 2 hours with Prime95 , no problems detected.
-Tested RAM with windows test, no problems.
-Tried different RAM sticks , GPU drivers still crash.
-Tried another PSU, didn't solve the problem.
-Tried different HDDs.
-Clean installed different windows systems to rule out driver's issues , same crash happened with windows7,8 & 10.
-Tried a different Graphic card & tested with Heaven Benchmark, crashes stopped.

What would cause a GPU to crash if not heat related problems? Should I attempt to open & clean the card? Reapply thermal paste? Bake the shit of my card ?
Right now , i am stuck with a 700$ paper weight. Any suggestions are welcome.

My build is:

PSU: Fractal Design Tesla 650W 80PLUS Gold R2
GPU: GIGABYTE GTX780 3 GB (384) of active 2xD H DP D5 OC
CPU: Intel Core i5-4570 BOX (3.2 GHz, LGA1150, VGA)
Case: Fractal Design Define R4 Black Pearl
RAM: 8 GB of DDR3-1600MHz Kingston HyperX Blu 2x4GB XMP kit
MOBO: ASUS B85M-G
HDD: Seagate 2TB HDD Desktop 64 MB SATAIII 7200rpm 2RZ

Thank you in advance , will rate best answer.
 
Solution

Salma_75

Distinguished
Jan 2, 2013
30
0
18,530
Not the kind of response i was hoping for :( , But sadly it seems to be the only thing i can do right now.
Lesson learned though ... this is the last time i buy anything from Gigabyte , their international warranty policies suck. I purchased this card assuming it's under 3 years warranty , but once i actually needed the warranty Gigabyte told me that my warranty is only 1 year, not 3.

Before this piece of crap, I had a 200$ Radeon card that i abused for over 4 years and it never crashed , even though i tossed it across the room once.
 

sunzone

Honorable
Mar 19, 2016
2
0
10,510
My knowledge is limited about this sort of stuff but, could you try downgrading your graphic card drivers to a older version and see if you continue to get the issue?(doesn't need to be a super old version, just a previous version.)
 

Salma_75

Distinguished
Jan 2, 2013
30
0
18,530

I tried that , Didn't work :(
 

sunzone

Honorable
Mar 19, 2016
2
0
10,510
Does this happen on all games? and what games does it not happen in?

Also, try in Nvidia Control Panel Setting-
Power Management Mode = Prefer maximum performance
Vertical Sync= OFF
This can help in some games but not all.
 

Salma_75

Distinguished
Jan 2, 2013
30
0
18,530


It happen on any game that is more demanding than a simple adventure "point & click" adventure.
List of fgames i tried:
world of warcraft
Minecraft
Witcher 2
Witcher 1
Batman arkham city
Alan wake
Goat sim
All of these games cause the display driver to crash, It is a hardware problem with my Card, i know that. I was just wondering if ,at this point, there is anything i can do other than tossing the card in the garbage. I do intend on taking the card apart & cleaning , might even try this reflow technique ...Nothing to lose.
 

maxalge

Champion
Ambassador


install msi afterburner, increase power limit to 115% and see if that helps



you can also try to slowly lower core clocks and see if that helps you reach stability
 
  • Like
Reactions: vairoxe
Solution

Salma_75

Distinguished
Jan 2, 2013
30
0
18,530


Well..... This actually worked!!!!!! Thank you so much!!
I was experimenting on World of warcraft since its a mid range game when it comes to the load it puts on GPU.

I increased power by 5% , nothing happened. Game crashed on load screen.
Lowered core clock by 5% , Game actually loaded !!! but crashed when game world was fully loaded.
Lowered by 18% , game didnt crash , i was able to play on lowest video settings.
-30% , was able to play the game on mid graphics settings.
-40% , game is playable on high graphics , 1080 x 1920 res & fullscreen with 100FPS.

What does that mean? Can i expect this stability to be maintained for the next couple of months? or is the card going to deteriorate with time again?

Thanks again man though, feels good to get any kind of positive results , even if they are temporary.

 

maxalge

Champion
Ambassador




it means either the power supply is not capable of properly powering the card or

the card had too low voltage for the core clocks it was running


since you say you tested with another power supply we can assume the latter


i suggest you slowly increase voltage and then core clocks until you are stable at 100% core clock while gaming

make sure to keep an eye on temps


you can easily test using something like the free benchmark valley, or heaven since you already have that



increase voltage and core clock a bit, run a benchmark program, if it is stable, increase core a bit more

any time the increase in core becomes unstable increase voltage a bit until it gets stable so on and so forth


if you can reach the proper full core speed of your card, and stay under 80 degrees while running valley then you are set




this should also help:


http://www.guru3d.com/articles_pages/gigabyte_geforce_gtx_780_ti_windforce_3x_review,27.html


 

Salma_75

Distinguished
Jan 2, 2013
30
0
18,530
[/quotemsg]



it means either the power supply is not capable of properly powering the card or

the card had too low voltage for the core clocks it was running


since you say you tested with another power supply we can assume the latter


i suggest you slowly increase voltage and then core clocks until you are stable at 100% core clock while gaming

make sure to keep an eye on temps


you can easily test using something like the free benchmark valley, or heaven since you already have that



increase voltage and core clock a bit, run a benchmark program, if it is stable, increase core a bit more

any time the increase in core becomes unstable increase voltage a bit until it gets stable so on and so forth


if you can reach the proper full core speed of your card, and stay under 80 degrees while running valley then you are set




this should also help:


http://www.guru3d.com/articles_pages/gigabyte_geforce_gtx_780_ti_windforce_3x_review,27.html


[/quotemsg]

What would happen if i left the core clock as is? At this point i would rather go for stability over performance, would leaving core clock settings at -40 affect the card on the long term?

also , since this is a factory OC version, is there a way to remove the factory OC settings?

 

maxalge

Champion
Ambassador


nothing bad, by lowering core clock you have in fact removed the OC
 

Zoli__

Commendable
May 23, 2016
16
0
1,510


Ive got a GTX 970 but it does the same problem and if its not it lockes the memory clock to 2x3005 mhz instead of 2x3500