• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Is My Graphics Card Dying?

Joined
Apr 27, 2021
Messages
2 (0.00/day)
Processor Intel Core i7-4790
Motherboard ASUSTek B85M-E
Memory Kingston 16GB (2 x 8GB) DDR3
Video Card(s) NVIDIA GeForce GTX 660
Software Windows 7
I've had my NVIDIA GeForce GTX 660 for about 5 years and it was working quite well for me. Today I was doing some video editing and noticed the image blacking out while I was doing color correction. Did a User Benchmark test (screenshot below) and noticed my GPU tested quite badly.

Now I'm wondering if there's no more hope for it, or if maybe there's something else going on in my PC that might be afecting it's performance.

I've also attached some screenshots from GPU-Z below. Can anyone give me some help?


ssssssss.png


ddd.gif

sadasdas.gif
 
Last edited:
USERBENCHMARK IS UNRELIABLE
i would recommend doing a test like unengine and comparing your results online or doing some of the tests in the tpu review https://www.techpowerup.com/review/msi-gtx-660-gaming/
and comparing results

also try dduing and see if that stops the blacking out
 
just for refrence this is the image i shard for reasons why userbenchmark is unreliable
1619481033262.png
 
Last edited:
a GPU does not degrade in "FPS" or Benchmark Scores.
i'd DDU the drivers in safemode and install them again without any internet connection.
 
a GPU does not degrade in "FPS" or Benchmark Scores.
i'd DDU the drivers in safemode and install them again without any internet connection.
why without internet?
ive never had to do that?
 
why without internet?
ive never had to do that?

Because windows 10 will try to grab drivers off their own server and install them automatically pretty much instantly if youre plugged in and device manager finds a new device that has no driver installed.
 
i
Because windows 10 will try to grab drivers off their own server and install them automatically pretty much instantly if youre plugged in and device manager finds a new device that has no driver installed.
have never had it do that
 
The thing about digital hardware is that it does not slow down as it ages; an old card will run the same exact speed as when it was new.

Now it could be the GPU solder joints are suffering from thermal fatigue.
 
Last edited:
Thats as far from the truth as it possibly can be.
Modern GPUs have a bunch of self-monitoring sensors; in general as cards age things like thermals get worse (paste degradation, dust buildup et cetera) which the GPU will pick up and generally will just clock itself lower as kind of a self-preservation feature.

Speaking of that, have you ever cleaned out your GPU and/or repasted it? At 5 years in, that may not be the worst idea.
 
Thats as far from the truth as it possibly can be.
Modern GPUs have a bunch of self-monitoring sensors; in general as cards age things like thermals get worse (paste degradation, dust buildup et cetera) which the GPU will pick up and generally will just clock itself lower as kind of a self-preservation feature.

Speaking of that, have you ever cleaned out your GPU and/or repasted it? At 5 years in, that may not be the worst idea.
True, the card will thermal throttle if it is overheating.
 
Actually it will drop boost clocks well before it like, thermally throttles in the actual traditional sense.
In any case, unmaintained old card = gets hot(ter) = decreased performance
So do some basic maintenance and see if it doesnt fix any of the issues you're encountering.
 
Thats as far from the truth as it possibly can be.
thermal throttling is not a silicon degradation.

repasting and cleaning /checking temperatures is another thing.
a "Dying GPU" does not lose performance.
 
a GPU does not degrade in "FPS" or Benchmark Scores.
i'd DDU the drivers in safemode and install them again without any internet connection.

sure it can. all chips and various other components on a GPU can degrade, some impacting performance more than others when they do degrade. most users likely will not see such problems, but they can certainly happen.


The thing about digital hardware is that it does not slow down as it ages; an old card will run the same exact speed as when it was new.

Now it could be the GPU solder joints are suffering from thermal fatigue.

are you and GerKNG drinking the same kool-aid or what lol? see my response above.


Actually it will drop boost clocks well before it like, thermally throttles in the actual traditional sense.
In any case, unmaintained old card = gets hot(ter) = decreased performance
So do some basic maintenance and see if it doesnt fix any of the issues you're encountering.

exactly, the more heat components have to endure the higher chance they will begin to degrade if not out right fail.
 
Yes, components degrade, but the performance remains the same till a threshold is passed and then errors are seen (throttling aside); that's the nature of digital.

For example, for TTL voltages

0-0.8V is low
above 2.0V is high

The chip might degrade from 0.1V to say 0.2V, but this will still be registered as low.
 
Last edited:
sure it can. all chips and various other components on a GPU can degrade, some impacting performance more than others when they do degrade. most users likely will not see such problems, but they can certainly happen.




are you and GerKNG drinking the same kool-aid or what lol? see my response above.




exactly, the more heat components have to endure the higher chance they will begin to degrade if not out right fail.
It's a machine. It will perform at the rated speeds until it fails completely in one way or another. It's not like anything biological where it would slow down because it's old. It'll degrade yes, but not in performance unless as previous said, thermal throttling.
 
It's a machine. It will perform at the rated speeds until it fails completely in one way or another. It's not like anything biological where it would slow down because it's old. It'll degrade yes, but not in performance unless as previous said, thermal throttling.
A car engine is a machine and the performance will degrade over time, even if properly maintained. Wear and tear still applies for PC components. How noticeable or big the drop is, that's a different story.
 
Last edited:
However... there always seems to be an however...

If it were an analogue machine like a car engine, it can indeed 'sag' and give out less power than when new.

The point here is analogue vs digital
 
A car engine is a machine and the performance will degrade over time, even if properly maintained. Wear and tear still applies for PC components.
That's more on moving parts and proper maintenance. You're not using a GTX580 as a piston. Different uses require different upkeep.
 
However, since I just sucessfully reflowed 2 video cards my mind is stuck on 'the black screen is due to thermal fatigue of the solder joints' :)
 
I am 95% sure this is due to overheating.
My GTX 650 Ti BOOSTs (basically the same card) blackscreen when they overheat. I repasted the cards with MX-4 (had original thermal paste), but forgot to reconnect the fans. Idle temps were 75C, and if I ran anything intensive the cards would get too hot and shut off.

OP, have you ever dusted/repasted the card?
 
I guess the thermal throttling mechanism doesn't work so well on these cards.
 
I guess the thermal throttling mechanism doesn't work so well on these cards.
Kepler cards only know how to run (blackscreen) not fight (downclock) lol
 
Because windows 10 will try to grab drivers off their own server and install them automatically pretty much instantly if youre plugged in and device manager finds a new device that has no driver installed.
This. If I need a clean driver install, I'll always unplug my WLAN dongle because Win10 tries to put older drivers to a card.
 
Back
Top