Please help me

[Swegly]

I can achieve high and stable overclocks with the my GTX970 G1 however even applying the slightest overclock or none at all causes my game to crash and this error to occur - 'Display driver has stopped and successfully recovered'

 

I have chatted with Microsoft, Nvidia, even corsair about this problem because I have absolutely no idea what to do about it. I have been dealing with it for weeks and gigabyte won't even reply to my ticket. I have reseated the card, my ram, ran tons of benchmarks and stress tests to find the error, repaired corrupted files, etc. I have tried every solution any company has recommended to me but nothing. 

 

The card is very well cooled and does not overheat, both the processor and ram are also working properly. Everything is updated. I have absolutely no idea where to turn.

 

If someone could help me I would really appreciate it I have been dealing with this for way too long and havent been able to use my PC.

[Volbet]

Well, your drivers crashing is usually a sign of an unstable overclock. How do you stress test the card, and are you sure the overclock is applied in that stress test?

[W-L]

-SNIP-

 

From initial outlook seems like you got a dud for overclocking, but what are your settings for your overclock?

[Swegly]

Well, your drivers crashing is usually a sign of an unstable overclock. How do you stress test the card, and are you sure the overclock is applied in that stress test?

 

From initial outlook seems like you got a dud for overclocking, but what are your settings for your overclock?

I have the latest windows 10 nvidia driver that includes the hotfix specifically designed for the TDR issues. The overclock can go much higher but it even crashes at +100 core and +250 mem and 110% power. no voltage changes. 

 

Edit: Im sure the overclock is applied when running tests

[Volbet]

I have the latest windows 10 nvidia driver that includes the hotfix specifically designed for the TDR issues. The overclock can go much higher but it even crashes at +100 core and +250 mem and 110% power. no voltage changes. 

 

Edit: Im sure the overclock is applied when running tests

What stress test(s) are you running? Not all stress tests are created equal. 

That is usually why I recommend using games as stress tests. Atleast they all stress somehwat equally.

[W-L]

I have the latest windows 10 nvidia driver that includes the hotfix specifically designed for the TDR issues. The overclock can go much higher but it even crashes at +100 core and +250 mem and 110% power. no voltage changes. 

 

Edit: Im sure the overclock is applied when running tests

 

Add some voltage and try increase slightly by +10-15Mhz each time on the core and stress test. You can just slide the voltage to the max on the Nvidia cards they have a physical cap on voltage so you can't overdo it on your cards nowadays.

[Swegly]

What stress test(s) are you running? Not all stress tests are created equal. 

Am using MSI kombuster 3 for stressing and unigine heaven for benchmarking. These programs do not crash and I get exceptional scores without and delay or artifacting. The crashing only occurs when playing steam gamess

[Swegly]

Add some voltage and try increase slightly by +10-15Mhz each time on the core and stress test. You can just slide the voltage to the max on the Nvidia cards they have a physical cap on voltage so you can't overdo it on your cards nowadays.

Can a lack of voltage be the cause of this issue? Other people who have the same card dont even touch their voltage slider until they hit +150 core.

[W-L]

Can a lack of voltage be the cause of this issue? Other people who have the same card dont even touch their voltage slider until they hit +150 core.

 

It all depends on your card hence the silicon lottery, I personally have to max my voltage on my G1970 to get a good overclock of +150 on the core.

[Volbet]

Am using MSI kombuster 3 for stressing and unigine heaven for benchmarking. These programs do not crash and I get exceptional scores without and delay or artifacting. The crashing only occurs when playing steam gamess

I can't recommend using Kompuster/FurMark for stress testing, seeing how both AMD and NVidia has made limitations in their driver when it comes to that program. 

It might have been fixed, seeing how I'm apperantly two revisions behind on Kombuster. 

 

Have you tried leaving Unigin Heaven on a loop, to check if it crashes the drivers aswell?

 

 

Can a lack of voltage be the cause of this issue? Other people who have the same card dont even touch their voltage slider until they hit +150 core.

Not all cards overclock the same. The sillicon lottery will determin your fate in that regard.

And yes, adding voltage will help stabilizethe overclock, but the card will produce more heat.

[Swegly]

It all depends on your card hence the silicon lottery, I personally have to max my voltage on my G1970 to get a good overclock of +150 on the core.

I just find it strange that the stress tests and benchmarks dont crash even without the voltage increased however games do. Is there any explanation for that?

 

Sorry if I come across as an amateur overclocker, it's because I am

[Swegly]

I can't recommend using Kompuster/FurMark for stress testing, seeing how both AMD and NVidia has made limitations in their driver when it comes to that program. 

It might have been fixed, seeing how I'm apperantly two revisions behind on Kombuster. 

 

Have you tried leaving Unigin Heaven on a loop, to check if it crashes the drivers aswell?

 

 

Not all cards overclock the same. The sillicon lottery will determin your fate in that regard.

And yes, adding voltage will help stabilizethe overclock, but the card will produce more heat.

So what clock and voltage should I start at and in what increments should I increase it. Is it normal to see some crashing at +100 core because I honestly have no idea whether that is minor or pushing the limits.

[Volbet]

So what clock and voltage should I start at and in what increments should I increase it. Is it normal to see some crashing at +100 core because I honestly have no idea whether that is minor or pushing the limits.

100 MHz is a pretty big jump to start off on. I usually recommend 10MHz as the biggest increments one should use. 

And everytime you jump up 10MHz, you stresstest with Unigin Heaven on a loop for 20-30 minuts to see if it's stable. 

Then you run a graphic heavy game to finalize the stresstes. You run this game for about as long as the Unigjn Heaven loop.

[W-L]

I just find it strange that the stress tests and benchmarks dont crash even without the voltage increased however games do. Is there any explanation for that?

Sorry if I come across as an amateur overclocker, it's because I am

 

I've had that happen I ran a benchmark for 2-3 hours with no issues on even higher core frequencies but it crashes about an hour into gaming, the loads created by benchmarking tools aren't as realistic as a real world gaming loads.

[Swegly]

I've had that happen I ran a benchmark for 2-3 hours with no issues on even higher core frequencies but it crashes about an hour into gaming, the loads created by benchmarking tools aren't as realistic as a real world gaming loads.

 

100 MHz is a pretty big jump to start off on. I usually recommend 10MHz as the biggest increments one should use. 

And everytime you jump up 10MHz, you stresstest with Unigin Heaven on a loop for 20-30 minuts to see if it's stable. 

I cannot tell you guys how helpful you are being to me right now, even more so than any other live chat person so I just want to say thanks for that.

 

What core clock should I begin with and what mV. Also after running a stress tester for ??? amount of time, how much should I increase each value? What about the mem clock?

[W-L]

I cannot tell you guys how helpful you are being to me right now, even more so than any other live chat person so I just want to say thanks for that.

What core clock should I begin with and what mV. Also after running a stress tester for ??? amount of time, how much should I increase each value? What about the mem clock?

 

Start at where your at say just a bit less +140 on the core and give it half of the max voltage around +40mV and see how much further you can push it, for stress testing I usually use the Unigine Valley Benchmark, say give it 20-30mins and try again to get it higher.

 

For the memeory I'd leave it where it's at for now and go for core first then do it again for the memeory. 

[Swegly]

Start at where your at say just a bit less +140 on the core and give it half of the max voltage around +40mV and see how much further you can push it, for stress testing I usually use the Unigine Valley Benchmark, say give it 20-30mins and try again to get it higher.

 

For the memeory I'd leave it where it's at for now and go for core first then do it again for the memeory. 

Is this a good way to do it - 

?

 

Also I guess before I was naive to think it was ok but this will definitely fix my problem. I have been dealing with it for weeks so you and Volbet made my day =)

[Volbet]

Is this a good way to do it - 

?

 

Also I guess before I was naive to think it was ok but this will definitely fix my problem. I have been dealing with it for weeks so you and Volbet made my day =)

It seems like a pretty good how-to guide. 

The only thing I would do different, is to not mix memory and GPU overclocking, as the video does to some extend. 

I would make an overclock on the GPU, with the desired speed and voltages, and then I would worry about the memory clock.

It makes the whole process less confusing, and it makes it easier to pinpoint where the instability might lie. 

[W-L]

-SNIP-

 

That's a way I guess to roughly dial things in to see where your at but you need to let it run the benchmark after you have it set to check stability. Usually it's recommend to take a look to see what others get and go below their settings by say 10% and go from there.

[Swegly]

That's a way I guess to roughly dial things in to see where your at but you need to let it run the benchmark after you have it set to check stability. Usually it's recommend to take a look to see what others get and go below their settings by say 10% and go from there.

 

It seems like a pretty good how-to guide. 

The only thing I would do different, is to not mix memory and GPU overclocking, as the video does to some extend. 

I would make an overclock on the GPU, with the desired speed and voltages, and then I would worry about the memory clock.

It makes the whole process less confusing, and it makes it easier to pinpoint where the instability might lie. 

I am going to start at +40 mV and +110 core. I will go ahead with the memory when I have this figure out. 

 

So I should bump the mV after raising the core clock to the max where i start seeing artifacts right. What is the max mV you would recommend?

[W-L]

I am going to start at +40 mV and +110 core. I will go ahead with the memory when I have this figure out. 

So I should bump the mV after raising the core clock to the max where i start seeing artifacts right. What is the max mV you would recommend?

 

Basically keep upping the core clock until it fails during a stress test, usually you get that driver error, then try again after raising the voltage. As said you can max it out to +87mV the voltage since it's limited physically my Nvidia so it's still going to be within safe ranges.

[Swegly]

Basically keep upping the core clock until it fails during a stress test, usually you get that driver error, then try again after raising the voltage. As said you can max it out to +87mV the voltage since it's limited physically my Nvidia so it's still going to be within safe ranges.

Is unigine heaven stressful enough? because i really like it.

[W-L]

Is unigine heaven stressful enough? because i really like it.

 

I would say yes, others also like to use Kombustor which also work well, it's mainly preference.

[Swegly]

I would say yes, others also like to use Kombustor which also work well, it's mainly preference.

I am not done yet but currently I am at +55 mV and +160 core... no artifacting or anything so far

[BuntzenB]

I can achieve high and stable overclocks with the my GTX970 G1 however even applying the slightest overclock or none at all causes my game to crash and this error to occur - 'Display driver has stopped and successfully recovered'

 

I have chatted with Microsoft, Nvidia, even corsair about this problem because I have absolutely no idea what to do about it. I have been dealing with it for weeks and gigabyte won't even reply to my ticket. I have reseated the card, my ram, ran tons of benchmarks and stress tests to find the error, repaired corrupted files, etc. I have tried every solution any company has recommended to me but nothing. 

 

The card is very well cooled and does not overheat, both the processor and ram are also working properly. Everything is updated. I have absolutely no idea where to turn.

 

If someone could help me I would really appreciate it I have been dealing with this for way too long and havent been able to use my PC.

I dont know if this is the problem at all but i had the same problem and i determined it was because my cpu was too weak to handle the oc and the game so what i did was limit the game to low priority or to only a couple of my cores and everything was fine and dandy. give it a shot.