Is my GPU going nuts?

[Cereal_Killer]

Howdy, folks! I haven't been here for quite a while, but I came back in dire need of help. You see, my Graphics Card, my MSI Radeon R9 270, bought just 3 months short of 3 years ago, appears to be going crazy. For two days now, I've been getting constant issues when playing my favourite game, Team Fortress 2. You see, when playing the game, the game would freeze for about two seconds, often with graphical artifacts appearing on screen, and then it would continue on, but with a frozen screen. I could hear the game progressing normally, I could walk around and interact with the game, but the screen was completely frozen. I asked for help on Steam Forums, and a person there suggested that I test my GPU with some software like 3DMark. Since 3DMark costs money that I don't have, I chose FurMark. And indeed, after reaching a certain temperature (usually around 54 degrees Celcius), the screen would get filled with horrid artifacts and glitches. Due to the relatively low temperature at which this issue starts happening, I don't think the card's overheating. I never overclocked it, so that's not the issue, either.

 

My question to you is: Does that mean that my card is dying and I need a replacement?

 

P.S.

Is "Howdy, folks!" a weird thing to come from a person who didn't grow up in southern USA?

[captain cactus]

Uninstall drivers with DDU? Replace thermal paste?

[WkdPaul]

Might be dead yes. Try to see if cleaning it out and changing the thermal paste helps. If you're still getting the same problem then it's probably dead.

 

You could try to undervolt and underclock it to see if it helps but I doubt it.

[Cereal_Killer]

1 minute ago, lots of unexplainable lag said:

Uninstall drivers with DDU? Replace thermal paste?

I used AMD Clean Uninstall Utility to get rid of old drivers and installed new ones. I didn't think of putting new thermal paste on it, but I don't have any around just now

[Praesi]

54° is no critical Temp at all. Check GPU-Z while gaming and see if the CoreClock is going  to idle or 0% when the Problems appear. I guess your Card is close to dead.

[Techstorm970]

Congrats on 100 posts! 

 

Edit: Aaaaaaannnnd this was my 400th post.  Freaky... 

[WkdPaul]

9 minutes ago, Cereal_Killer said:

I used AMD Clean Uninstall Utility to get rid of old drivers and installed new ones. I didn't think of putting new thermal paste on it, but I don't have any around just now

You can ask the local small PC repair shops for the one time use packs. They probably have some laying around or if they charge you it might be minimal.

[Techstorm970]

Also, I don't really know how to fix your problem.  Yes, it could be dying or effectively dead.  If it is dying, though, then I can help you with finding a replacement.  

[sof006]

On a scale of 1 to dusty how dusty is your PC?

 

Try taking out your GPU and reseating it. Maybe blow some compressed air into the PCIE slot (just to make sure its clean in there) and put GPU back in. Test again. Something similar happened with me about a year or two ago and I had a bit of dust obstructing my PCIE slot and it was causing some weird stuff to happen which made me think it was my card.... It wasn't obviously.

[Cereal_Killer]

I just noticed I forgot to mention an important part of my issue. About every 3rd of these freezes also causes my system to crash (I'm running Windows 8.1).

16 minutes ago, Praesi said:

Check GPU-Z while gaming and see if the CoreClock is going  to idle or 0% when the Problems appear. I guess your Card is close to dead.

I did what you said, and unfortunately my computer crashed a few seconds after the problem appeared, so I'm not sure if the information I'm about to give you is right from that moment or sometime earlier, but here we go anyway

Core Clock: 955.0

Memory Clock: 1400.0

Temperature: 52.0

Fan speed (%): 28

Fan speed (RPM):1300

GPU load: 50

 Memory Usage (Dedicated) [MB] 727

Memory Usage (Dynamic) [MB] 64

VDDC [V] 1.188

 

For the previous 5 minutes of recorded values, there are only slight variations in GPU load, Fan speed, Temperature, Memory Usage, etc. Nothing out of the ordinary

[Praesi]

4 minutes ago, Cereal_Killer said:

I just noticed I forgot to mention an important part of my issue. About every 3rd of these freezes also causes my system to crash (I'm running Windows 8.1).

I did what you said, and unfortunately my computer crashed a few seconds after the problem appeared, so I'm not sure if the information I'm about to give you is right from that moment or sometime earlier, but here we go anyway

Core Clock: 955.0

Memory Clock: 1400.0

Temperature: 52.0

Fan speed (%): 28

Fan speed (RPM):1300

GPU load: 50

 Memory Usage (Dedicated) [MB] 727

Memory Usage (Dynamic) [MB] 64

VDDC [V] 1.188

 

For the previous 5 minutes of recorded values, there are only slight variations in GPU load, Fan speed, Temperature, Memory Usage, etc. Nothing out of the ordinary

Repeat it. Run Valley Benchmark and observe with GPU-Z. If your GPU has Problems it will appear there.

[Cereal_Killer]

20 minutes ago, sof006 said:

On a scale of 1 to dusty how dusty is your PC?

 

Try taking out your GPU and reseating it. Maybe blow some compressed air into the PCIE slot (just to make sure its clean in there) and put GPU back in. Test again. Something similar happened with me about a year or two ago and I had a bit of dust obstructing my PCIE slot and it was causing some weird stuff to happen which made me think it was my card.... It wasn't obviously.

How dusty is my PC? If dusty is 10, then I'd say about 4, tops. Motherboard seems rather clean. There are some bits here and there on the CPU fan. A bit of buildup on the PSU and case bottom, but nothing special. GPU fans have noticeably more dust than other parts, but it still doesn't look like that much. I'll clean it with compressed air, of course, it's been a while since I did that, but it doesn't strike me as the direct cause of the issue, especially since it first happened two days ago and has been so frequent I didn't manage to finish a single match of TF2 since that day, whereas dust buildup is a slow and gradual process. I don't have a screwdriver at hand nor compressed air, so I can't do it now.

 

[Cereal_Killer]

12 minutes ago, Praesi said:

Repeat it. Run Valley Benchmark and observe with GPU-Z. If your GPU has Problems it will appear there.

Ran Unigine Valley, and at around 51 degres celcius at Extreme HD, it just crashed. The PC runs fine, but I'll run the program one more time, this time at just Extreme.

 

Okay, I ran it. It was in windowed mode so I could see other stuff on screen, which turned out to be pretty valuable. I barely ran it and already had the screen freeze with intense lense-flare-like spots on the screen, but after a second it recovered and played again. About 10 seconds later, the same thing happened, but more intense. Then there was just constant freeze, and I decided to cut it and close the program. What's interesting is that, every singe time the program froze, Windows showed me a notification saying "display driver stopped responding but recovered"

Here's the GPU-Z log from the test - GPU-Z Sensor Log valley.txt

What I noticed is that whenever the screen froze, the card jumped from 955 mhz core and 1400 mhz memory clocks to 300 mhz core and 150 mhz memory clocks, and instantly dropped to 0% usage, then after a while recovered and kept going at the proper 955/1400 clocks, only to be cut out again a few moments later.

[Praesi]

Exact the same behaviour like my nephew's gtx670. I reinstalled the drivers several times, changed cables, changed RAM and did everything to make sure nothing else was causing this. Unfortunately his Card is dead now completely and i think the same is happening to your Card now.