Periodic short freezes with looping audio, and sometimes TDR failure

[Roflmaster]

Specs:

• 
  ■ OS: Windows 7 64-bit ultimate
  ■ GPU: GTX 970
  ■ CPU: i5-2320 @ 3.30 GHz
  ■ Motherboard: Gigabyte P61-DS3-B3
  ■ PSU: XFX P1-600B-XTFR (600W)
  ■ RAM: 2 x 4GB DDR3 @ 1333MHz

I haven't had similar problems with this rig and I am sure that these components work well together since I ran my games smoothly before, on the same graphics settings.

These freezes happen mostly in video games, but every once in a while they will also happen while browsing the internet and streaming music through it.
Sometimes the freezes occur once a second and last only for a fraction of a second, while at other times the freezes happen maybe 2 or 3 times a minute and then last about half a second. During those times the game doesn't accept any input from the keyboard or the controller. Pressing alt+enter or changing from windowed to fullscreen (or vice versa) fixes the problem for a few seconds, but then it comes back.

If I am unlucky, the screen freezes for longer, and after around 2 seconds goes black and the monitor reports "no signal". When this happens though, the sound doesn't start looping (like it usually does), i can use VoIP and other people online can hear me; it seems like everything but graphics is functioning well. After those 2 seconds the audio finally starts looping and this situation doesn't resolve so I restart the computer by holding the power button. The same happens through HDMI. It seems that those 2 seconds are the TDR delay. Recently I changed the delay to 10 seconds in the registry in the hopes of giving the GPU more time to reset but I haven't had time to notice any differences.

Because the screen goes black I don't get a BSoD but Windows does create a crash minidump. The minidump sometimes reports VIDEO_TDR_ERROR and sometimes VIDEO_TDR_TIMEOUT_DETECTED, but on my end the crashes seem identical. (bug checks 0x116 and 0x117 respectively).

This is what msi afterburner logged while the freezes were happening: https://imgur.com/a/7zR3Q
The first example is while browsing, the second while playing a video game.
Notice how the framebuffer usage drops to 0% precisely every 30 seconds, while at the same time my GPU's clock frequency rises, causing stuttering. In the second example when FB usage is low, GPU usage varies wildly.

Things I already did to troubleshoot:

• 
  ■ Reinstalling drivers: using the DDU utility i uninstalled my nvidia drivers in safe mode and installed the newest ones, and later even some older version but it didn't help. NOTE: I didn't manually delete any folders and I didn't try eradicating all traces of nvidia from my pc so maybe the clean install of new drivers wasn't that clean.
  ■ Checking if I'm overheating: CPU temps always below 70, GPU temps always below 65
  ■ Ran furmark to test my GPU stability: it didn't crash but I didn't let it run for very long either.
  ■ Memtest to check my RAM: left overnight (10hrs), 0 errors found. I left both sticks in during the tests.
  ■ Since TDR errors pointed at maybe a faulty PSU i cleaned the fans of my PSU (they were very dusty and maybe it was overheating?). The freezes went away for a short time but came back, might have been a coincidence.

Things i didn't do:

• 
  ■ Reinstalling windows (seems like overkill to me)
  ■ Reseating all my components
  ■ Touching the PSU to see if it's overheating since I can't see its tempereature through software (I'm afraid of touching it)

Some observations:

• 
  ■ TDR failure usually happens when i alt+tab back into a game or when a new round begins in an online match (perhaps it happens when assets are loading?)
  ■ TDR failure and the whole system freezing into black srceen happened in more graphically demanding, recent games. These more demanding games, however, didn't suffer any temporary short video and audio freezes, only crashes to black screen. Less demanding, older games never crashed my computer outright, but they did suffer from freezing and stuttering. Even 2d games, although some games seem to be exempt.
  ■ Recording the screen with fraps didn't capture the audio stutter, and video doesn't freeze, the framerate just drops in the captured video. Why is there a discrepancy between what is recorded and what is displayed while it's actually happening? fraps example
  ■ Windows 7 updated my nvidia graphics drivers just before the problem started happening, this might only be a coincidence. I am unable to uninstall this update (perhaps because it's 3rd party software).

Hopefully someone can help me resolve this issue.

 

[TwilightRavens]

Do you have another GPU laying around or even one you can borrow from a buddy? Because it sound like your gpu is on its way out. But the only way to rule that out would be to use a different gpu or even integrated graphics.

 

[Roflmaster]

Do you have another GPU laying around or even one you can borrow from a buddy? Because it sound like your gpu is on its way out. But the only way to rule that out would be to use a different gpu or even integrated graphics.

My motherboard doesn't have any graphics ports so I can't use the integrated graphics. I might be able to get my hands on another GPU though.

My GPU is still under warranty, so can I request a replacement without being 100% sure that mine is dying? Or do I have to diagnose a faulty GPU myself before they give me a replacement?

 

[TwilightRavens]

If you do try another card and it works without a hitch then you will have pretty much diagnosed that the card is dying. And usually they will test the card when they get it to see if it is indeed at fault. I say if its under warranty go ahead and RMA it, but only after you try another card first.

-edit: Also try moving it down a PCI-E slot if you have another, theres a chance the slot could be going bad.

 

[Roflmaster]

I'm reviving this because I still have the same problem.

I RMA'd the card and installed an older GPU into my system. I have not experienced any issues with this older card.
My GTX 970 came back and they have deemed it to be not faulty. They say that they have "tested it for 48 hrs under 3d load" and that it functions normally.

For the next month or two I didn't have any problems with the GTX 970, but then I started getting crashes again. It even happened while browsing the web and loading gifs and short audio loops and freezes could be heard when switching tabs while music was playing in one of them.

Perhaps it has to do with VRAM? I have hardware acceleration enabled in firefox so freezing when switching tabs is probably due to the GPU moving its memory around? The crashes also frequently happen when the video games are loading, i.e. when I'm switching levels. Another thing that points to this is that when the game is only freezing up but not crashing, the framebuffer usage drops a lot (and GPU usage goes all over the place).
These are just my guesses since I don't actually know much about graphics cards

What are the next troubleshooting steps?

-edit: I only have one PCI-E slot unfortunately, and the other card wasn't acting up so I guess the slot is probably fine?

 

[TwilightRavens]

I'm reviving this because I still have the same problem.

I RMA'd the card and installed an older GPU into my system. I have not experienced any issues with this older card.
My GTX 970 came back and they have deemed it to be not faulty. They say that they have "tested it for 48 hrs under 3d load" and that it functions normally.

For the next month or two I didn't have any problems with the GTX 970, but then I started getting crashes again. It even happened while browsing the web and loading gifs and short audio loops and freezes could be heard when switching tabs while music was playing in one of them.

Perhaps it has to do with VRAM? I have hardware acceleration enabled in firefox so freezing when switching tabs is probably due to the GPU moving its memory around? The crashes also frequently happen when the video games are loading, i.e. when I'm switching levels. Another thing that points to this is that when the game is only freezing up but not crashing, the framebuffer usage drops a lot (and GPU usage goes all over the place).
These are just my guesses since I don't actually know much about graphics cards

What are the next troubleshooting steps?

-edit: I only have one PCI-E slot unfortunately, and the other card wasn't acting up so I guess the slot is probably fine?

Have you tried a clean install of drivers? Like for example using nvidia's cleanup tool, then the Display Driver Uninstaller tool? Also did you start noticing it after a driver update, if so go back to the last one you have working as sometimes driver releases can just botch a card, or if its an older driver try the most recent which at the time of me posting this is 390.77 I believe.