Overclocking/declocking issue

[Thedefh]

Hello,
a friend of mine gave me his 980 (GV-N980D5-4GD-B) which has been crashing in games for the last year,
the thing is my friend has been tempering with the clocks etc and from what I can tell now after taking the time to review the specs he has permanently altered all the settings and now there is no option to go back.
does anyone here has the stock firmware so I could try and flash it somehow (if any of you have a tutorial it would be helpful) or any idea how i tweak the settings back to the original specs?

[Jurrunio]

4 minutes ago, Thedefh said:

he has permanently altered all the settings and now there is no option to go back.

Unless he flashed the card's BIOS (which is unlikely, most dont know how to do it), all altered settings only exist in the OC software he used. Uninstall that and things will go back to normal

 

Otherwise it means the card isnt stable at factory settings, which is another level of problem.

[Thedefh]

It's not on his computer anymore,it's on mine.
 Up until now I didn't have any overclocking utilities which I installed when I saw I get much higher clocks than I should which I guess is the cause of the game crashes (also extremely high temperatures)  and the settings aren't consistent with the factory ones

[Jurrunio]

1 minute ago, Thedefh said:

I saw I get much higher clocks than I should

It's more surprising if you only get the base or boost clocks stated by the manufacturer. Starting from 900 series, "GPU boost" (a name by Nvidia, could search that up yourself) lets the GPU speed up beyond spec depending on power, voltage and thermal headrooms.

 

5 minutes ago, Thedefh said:

(also extremely high temperatures)  

How high? Could be just dried out thermal paste, happens on even newer cards.

[Thedefh]

it gets to 90 degrees Celsius in less than a minute in when doing the 1080P stress test in furmark (my 970 gets to about 70 max in the same test)

and I have already switched out the thermal paste as I assumed that it could be the cause of the game crashes.

[Jurrunio]

Mounting pressure? Or the cooler could be faulty in the first place. There's a reason why blower card is usually not recommended, they dont have the headroom to get worse.

 

Also dont run Furmark, prone to blowing up cards especially Nvidia reference cards before Turing.

[Thedefh]

"Mounting pressure"? I'm not familiar with the term or the way to measure it.
About the cooler - When the card was new it was OK with no crashes it's been 5-6 years since my friend bought it.
If you don't recommend Furmark - what do you recommend? 
 

[Jurrunio]

1 minute ago, Thedefh said:

"Mounting pressure"? I'm not familiar with the term or the way to measure it.

The pressure of heatsink pressing down on the GPU. Could be affected by the torque you applied onto the 4 main screws

 

2 minutes ago, Thedefh said:

If you don't recommend Furmark - what do you recommend? 

A game that can GPU bottleneck your system

[191x7]

FurMark can not damage a properly working card. It can damage only faulty cards (even new if they have a fault or a bad VRM design).

 

If the card is properly cooled, FurMark won't kill it with temperature.

If the VRM-s have enough phases, are cooled sufficiently, and do not have a factory defect or do not suffer from extremely old age - the VRM won't blow no matter what load you put on it.

 

FurMark simply puts a huge load on the graphics card. Not even the max load since it's OpenGL based.

If it fries your card, it did not fry a good card, it killed off a faulty device.

[Jurrunio]

6 minutes ago, 191x7 said:

FurMark can not damage a properly working card. It can damage only faulty cards (even new if they have a fault or a bad VRM design).

 

If the card is properly cooled, FurMark won't kill it with temperature.

If the VRM-s have enough phases, are cooled sufficiently, and do not have a factory defect or do not suffer from extremely old age - the VRM won't blow no matter what load you put on it.

 

FurMark simply puts a huge load on the graphics card. Not even the max load since it's OpenGL based.

If it fries your card, it did not fry a good card, it killed off a faulty device.

Nvidia reference cards were long plauged with poor VRM designs and barely big enough inductors... Espeically the memory VRM, bad news since Furmark stresses memory especially hard, much harder than even games.

[191x7]

5 minutes ago, Jurrunio said:

Nvidia reference cards were long plauged with poor VRM designs and barely big enough inductors... Espeically the memory VRM, bad news since Furmark stresses memory especially hard, much harder than even games.

Even Linus said there's no "game stable", it's either stable or unstable. If it crashes any stress test after any amount of time, it's not stable.

 

If you must have a ligher load, people like Unigine benchmarks like Heaven or Valley.

[Thedefh]

I seem to have another inconsistency:
I see this model instead of the one etched to the card itself

[Jurrunio]

9 hours ago, 191x7 said:

Even Linus said

how much does Linus know anyway

 

9 hours ago, 191x7 said:

there's no "game stable", it's either stable or unstable. If it crashes any stress test after any amount of time, it's not stable.

There is. Not so commonly seen in GPUs since most systems are GPU bottlenecked, but it happens on both CPUs and GPUs. Otherwise you won't hear anyone talk about "settings that only posts" or "settings that can only complete Cinebench R20". They can do it over and over again, just not enough for other tasks.

 

8 hours ago, Thedefh said:

I seem to have another inconsistency:
I see this model instead of the one etched to the card itself

The card's cooler is the same one as on the reference card right? You could check if base and boost clocks of the card matches up with the reference card specs or not.