Kernel Drive Issues

[Jean Ribault]

I had a very similar graphics issue with a separate Steam game.  I have a high end system as do I'm sure most of you, with Crossfire available for the two cards.

 

Symptoms were a random lockup of the computer system.  Nothing but a system reset would recover it, so total loss of game.  There was no problem with Windows.

 

The solution after weeks (to months?) of attempting different fixes, was simple.  Had to reduce the overdrive settings on the two video cards.  There was an optimal setting I came to that completely eliminated these crashing issues, and I made no other system changes.  I had to make the change to both the cards since that game played in Crossfire mode.  They were AMD Radeon HD cards, like you mentioned above, and there were no cooling problems with the system, as has been suggested above as possible cause - I monitor all system temps.  It was clearly the overdrive settings, all things considered, in my case.

 

Maybe that will help, you might try adjusting some of your settings in Catalyst software to see if it helps/resolves your issues?

 

 

 

EDIT: I missed this, but now see that Ink mentioned this exact solution above.  Worked for me.  

Edited April 13, 2016 by Jean Ribault

[LordAlexander]

Well another night with no crashes.   I am going to say Turning off FXAA fixed my crashing problem.  It does not cause me to crash in other games but in this one it seems to.

[CHVNGE]

Generally an error saying "Window Nvidia kernel version xxx has stopped responded and has restarted" means that the graphics card has for some reason hung and has not sent and/or calculated a process for longer than 2 seconds. 2 seconds being the default time the OS gives for the GPU to send information to the output. If information takes any longer than 2 seconds then this error appears.

 

You can do a simple registry edit where we completely disable this error messaging.. but instead of getting an error and the driver restart you will get a system freeze forcing you to hard power off your system where the driver did not restart. We don't achieve anything by editing the TdrDelay or disabling the TdrLevel.

 

People don't like being told they have a faulty motherboard or need to replace system ram but these two factors are very legitimate reasons why you might be getting these problems.

 

If you are lucky.. lowering the gpu clockspeed will fix your problem. It seems that gpu load can max out causing a crash with overclocked or factory overclocked cards and also cards with certain boost capabilities.

 

Another problem could be due to a gpu bios bug. Some people have had to flash their gpu bios to fix this problem but I must stress that if you don't know what your doing then do not try to attempt this fix.. at least not until you are fully confident knowing you will not make a mistake.

 

Known theories include observations where clock speed boost exceeds the clock speed limit causing the card to momentarily hit a ceiling it is not comfortable with.

Flashing your bios can bring the numbers back into reasonable parameters if done properly.

Some people suggest that after updating Nvidia drivers is the moment when they notice they start to get these problems with their bios. There are also reports where people have a faulty bios after trying to setup an SLI configuration where one card for some reason has a faulty bios after installing an update. The problem with this theory is that we are not 100% certain of the state of the bios before the problem appears so its kind of difficult to imagine Nvidia maliciously altering the bios through an update where the parameters for variables are incorrect for the cards capabilities.

 

Its a pretty weird occurrence because we don't know for sure what is causing a recently working gpu to have such problems.. I have done a bunch of testing lately with possibly close to a hundred or more crash reports being sent to the guys here at gamelabs. And from what I have seen so far with regards to the crashes is that I get is some kind of buffering issue where the gpu takes to long to process.

 

I tried to use the game setting vsync and frame limiter but this caused instant crashing for me and this was very difficult to reset because we cant adjust the game setting before loading into open world causing me to repeatedly crash upon loading up.

 

I have not yet flashed my bios as I am waiting on some system ram that I will be lending from a friend to do one final test before I go a long and flash the bios. 

 

It is not a game issue, it is not a game patch issue.. regardless what Nvidia support have told you this issue is normally always caused by a bugged gpu bios, degraded ram or faulty motherboard and rarely anything to do with overheating or reinstalling drivers.. which would explain the low temps and continued error after downloading a multitude of different drivers.

[LordAlexander]

I have had my driver crash again but now it is down to 1 time in a week vs 3 times a night.  I stress again that I took a week off of this game and played other games on max settings and never had my video driver crash in those games.   turning off FXAA helped a lot.