Fluctuating frame rate and not sure what to do

[TheoneandonlyMrK]

Replace thermal compound on the gpu die, and give your case and fans, to included the videocard heatsink and fan a good dust removal.

Replace thermal compound on cpu too,

I'll give everything a good dust service tomorrow but I have to be honest, I've never touched the thermal paste on the card so I'm not sure why I would need to replace that. The card has been fine for over 2 years now. Its never been overclocked so it's never run hotter than it should so I doubt the thermal paste needs replacing.

I'm more inclined to believe that maybe the PSU us on is way out so I will check tomorrow if I have another port on the PSU for a second card that has never been in use to see if it will change things.

Is that an acceptable plan?

Most modern cards run hot and two years is ample time to degrade thermal interface Paste, he's also giving a cheaper option then a new PSU

 

[Wolvyreen]

@theoneandonlymrk As I said already a few times, I'm not against changing the paste... I'm only against it as a first approach. Not to mention my problems started right after a power cut... I'm sure the power cut didn't fry my paste

 

[phanbuey]

plus it's fluctuating memory clock, has nothing to do with the core temp on the card...

If it was overheating only the core would decrease and it would sound like a jet.

 

[Wolvyreen]

plus it's fluctuating memory clock, has nothing to do with the core temp on the card...

Also, if you look at the GPUz logs you will see the temps never exceed 60 degrees even when gaming. Hence why thermal paste is the last resort

 

[TheoneandonlyMrK]

As I said already a few times, I'm not against changing the paste... I'm only against it as a first approach. Not to mention my problems started right after a power cut... I'm sure the power cut didn't fry my paste

So if I must explicitly explain how that's wrong.

You use the card two years at stock , the chip , Tim and actual heatsink degrade over time becoming less effective at their purpose but still within a working range.
A power spike comes along and is largely suppressed by the PSU etc, but the exaggerated ripple seen by the card pushes the chips degradation to a point where it's not happy anymore at a given temp /clock and power limit so it's downclocking itself until it cools then clocks up.

Rejuvinating the cards cooling capacity can fix that, and certainly has for a few PC's i have fixed.

Your PSU likely would have crashed the pc due to current limit if you had tried Those Gpu settings on day 1.

Could also be the psu given your replys while I was writing

 

[phanbuey]

Yeah if it's not software its then PSU.... IMO the thermal paste is fine, no reason to fiddle with it.

 

[Wolvyreen]

So if I must explicitly explain how that's wrong.

You use the card two years at stock , the chip , Tim and actual heatsink degrade over time becoming less effective at their purpose but still within a working range.
A power spike comes along and is largely suppressed by the PSU etc, but the exaggerated ripple seen by the card pushes the chips degradation to a point where it's not happy anymore at a given temp /clock and power limit so it's downclocking itself until it cools then clocks up.

Rejuvinating the cards cooling capacity can fix that, and certainly has for a few PC's i have fixed.

Your PSU likely would have crashed the pc due to current limit if you had tried Those Gpu settings on day 1.

Could also be the psu given your replys while I was writing

Today is day 1 after the power cut. It was mentioned in the first post that the PC froze and I had to reboot. Hmmm

 

[TheoneandonlyMrK]

Today is day 1 after the power cut. It was mentioned in the first post that the PC froze and I had to reboot. Hmmm

I am not sure if the cut even matters you had issues before it?

And you said you had not overclocked it before.
And this first attempt at a power slider nudge then ended in a hard lock, crash.

So like I said it might have crashed anyway if you had tried those gpu settings two years ago, but anyway I agreed above it could be the PSU and im not here to argue ,i was just trying to help.

Have a nice weekend and good luck.

 

[Wolvyreen]

I am not sure if the cut even matters you had issues before it?

And you said you had not overclocked it before.
And this first attempt at a power slider nudge then ended in a hard lock, crash.

So like I said it might have crashed anyway if you had tried those gpu settings two years ago, but anyway I agreed above it could be the PSU and im not here to argue ,i was just trying to help.

Have a nice weekend and good luck.

I really do appreciate the help but I still feel that the fact that the temps are well within acceptable limits during gaming should immediately steer me clear from taking the card apart to replace the thermal paste as a first stab at solving this.

 

[trog100]

you may have a failing psu or a failing card.. its not running anywhere near full power for reasons as of yet unknown..

the options are start swapping out parts.. first try another gpu..

trog

 

[phanbuey]

well a new PSU is $80 and a new GPU is $200... and if the card was fried it wouldn't intelligently fluctuate clocks up and down. It would most likely die while vomiting artifacts.

If anything, check software first... then new psu... and then new GPU.

 

[Wolvyreen]

@trog100 I wish I had another one to test it with. Also wish I had another PSU.

 

[phanbuey]

if software doesn't work then something like:

https://www.amazon.com/dp/B01B72W1VA/ref=twister_B07JJ8BSMZ?_encoding=UTF8&psc=1

 

[agent_x007]

Did card ever had a hard moded removed power limit (like liquid metal on shunts) ?
Cause when I see green bar on GPU-z "Perf. limit" I say you are bouncing off power limit.
Your low temps are result of too low power limit that was put in place (by something or someone).

 

[Wolvyreen]

Did card ever had a hard moded removed power limit (like liquid metal on shunts) ?
Cause when I see green bar on GPU-z "Perf. limit" I say you are bouncing off power limit.
Your low temps are result of power limit that was put in place (by something).

I don't like over clocking or modding so I've never made any changes to this card at all. Still stock out of the box

if software doesn't work then something like:

https://www.amazon.com/dp/B01B72W1VA/ref=twister_B07JJ8BSMZ?_encoding=UTF8&psc=1

Think I will have to consider getting A PSU to test it

 

[agent_x007]

What maximum TDP/Power do you see in GPU-z ?
Also, what is your maximum GPU Voltage ?

 

[Wolvyreen]

What maximum TDP/Power do you see in GPU-z ?
Also, what is your maximum GPU Voltage ?

The GPUz logs are in the first post that should show you what you're are asking

 

[rtwjunkie]

@trog100 I wish I had another one to test it with. Also wish I had another PSU.

Can you think of anyone you could borrow from? Or perhaps anyone else with a PC that you might try your card out in? If the card works fine there, we can eliminate the card as the problem.

 

[Wolvyreen]

I'll need to ask some of my buddies but none of them are enthusiasts. So I don't think they would have something that would be able to stand up to the 980ti

 

[agent_x007]

I think you got a damaged power monitoring circuit (you get power limited when card isn't maxing it according to logs).

In short :
Under 200W power + around 50% TDP = Power limit reached (according to logs).

 

[Wolvyreen]

I think you got a damaged power monitoring circuit (you get power limited when card isn't maxing it according to logs).

Would this be the PSU??

 

[agent_x007]

If 6-pin/8-pin plug doesn't get the power it needs... then maybe ?
(PSU's shouldn't be inteligent enough to power throttle a GPU though, especially when all plugs are connected - they either work, or shut down/crash the PC)

You could mod vBIOS and see how card behaves when it's undervolted, or has software power limits removed.

 

[Wolvyreen]

I took a video this afternoon that I planned on uploading tomorrow but I did it now to show you what's happening.

 

[trog100]

@trog100 I wish I had another one to test it with. Also wish I had another PSU.

that is where the guy in the repair shop has the advantage over you... he always has some swap-able parts.. something i used to call the repair mans secret..

trog

 

[Mayclore]

Hold on.

What power supply do you actually have? Also, what version of Windows? Next, I'd be curious to know what a monitoring program such as HWiNFO64 seems to think your 12V voltage is. If it's wildly out of spec, then yes, your unit is probably borked. If not, something else is at play.