GPU works for a few seconds then stops working

[Galaxyz]

Yesterday my GPU suddenly gave me a black screen. After restart I was having this issue where it displays BIOS/POST and Windows login screen for about 10-20 seconds then goes back to a black screen. This seems to have nothing to do with the type of content displayed. Going into my BIOS, logging into windows, just doing nothing, etc. None of these prevent it from going black.

I have tried uninstalling drivers, same thing happens except this time with an 800x600 resolution rather than 1920x1080.
I have an ASUS STRIX GTX 970 so fans are an unreliable way to tell what's going on (semi-passive fans).

iGPU is automatically turned off with GTX 970 in PCIe slot (recognizes GPU). When I pull out the 970 iGPU takes over and works fine (I'm using this PC with my Intel HD 4600 now).

Also tried another dedicated GPU (very old GPU, so an issue with the 8-pin connector cannot be confirmed), everything works like it's supposed to.

This system has been successfully running for years and I'm not sure what could trigger the GPU to die now (if you can even call it that, it's clinging to life), or how to fix it.

Any help is appreciated.

 

My specs:

- 450W PSU (80+ Gold)

- ASRock B85 Fatal1ty Killer

- i5 4460

- 16 GB RAM

- And a bunch of SSD/HDDs

[191x7]

Is the monitor always plugged into the integrated card?

[Galaxyz]

Just now, 191x7 said:

Is the monitor always plugged into the integrated card?

Nope, have just been using this to test what was wrong. It correctly turns off whenever the 970 is inserted and on when it's not. 

[k.m.p]

is the gpu clean or dusty?

excess dust buildup can become conductive and start frying the gpu but only if its excessive.

have you got a different pc or another gpu to test with?

[Galaxyz]

57 minutes ago, k.m.p said:

is the gpu clean or dusty?

excess dust buildup can become conductive and start frying the gpu but only if its excessive.

have you got a different pc or another gpu to test with?

Completely cleaned it today although my computer is never excessively dusty to begin with. I still have to find someone with a beefy enough PC to test my card on. 

I have tried a couple more things by now such as resetting the CMOS, with no positive results so far.

[k.m.p]

20 minutes ago, Galaxyz said:

Completely cleaned it today although my computer is never excessively dusty to begin with. I still have to find someone with a beefy enough PC to test my card on. 

I have tried a couple more things by now such as resetting the CMOS, with no positive results so far.

unfortunately it does seem like the card is fried, as its a 970 I'm assuming its now well out of warranty,Im in the same boat, fans on my gtx 980 also died a few months back, not really got time to play games right now so gonna wait until October to rebuild my pc.

Edited March 8, 2019 by k.m.p

[Galaxyz]

5 minutes ago, k.m.p said:

unfortunately it does seem like the card is fried, as its a 970 I'm assuming its now well out of warranty,Im in the same boat, fans on my gtx 980 also died a few months back, not really got time to play games right now so gonna wait until October to rebuild my pc.

A good few seconds of signal + spinning fans followed by a black screen doesn't sound quite that fried to me... This behavior is the most confusing part of it all to me. I'm definitely considering getting a new one but I'd much rather resurrect this one.

[k.m.p]

2 minutes ago, Galaxyz said:

A good few seconds of signal + spinning fans followed by a black screen doesn't sound quite that fried to me... This behavior is the most confusing part of it all to me. I'm definitely considering getting a new one but I'd much rather resurrect this one.

sorry, missed that part, in the few seconds of signal are you able to see the bios?

if you are, what are your gpu temps?

could be the thermal paste application.

It seems like all other problems have been ruled out and you've basically tried everything else you can.

I doubt its the powersupply since 450w should be more then enough for that setup but not out of the question.

[Galaxyz]

1 hour ago, k.m.p said:

sorry, missed that part, in the few seconds of signal are you able to see the bios?

if you are, what are your gpu temps?

could be the thermal paste application.

It seems like all other problems have been ruled out and you've basically tried everything else you can.

I doubt its the powersupply since 450w should be more then enough for that setup but not out of the question.

Good point, it's very hard to make it far enough into the BIOS in that time span but I could try enabling the iGPU as primary and seeing if I can monitor temps like that. Parts have been working together for a while with no changes applied so I doubt that the PSU all of the sudden isn't enough anymore (although I must admit it's on the low side).

EDIT: It's impossible to get to my GPU temps using my iGPU or going into my BIOS fast. Though my motherboard and CPU temperatures seemed to be somewhere around 30-40C and touching the card right after one of these black screens felt cold.

Edited March 8, 2019 by Galaxyz

[Galaxyz]

In the time that has passed since making this thread I have discovered a few things.
1. I have a new computer now and the 970 does the exact same thing on my new PC.

2. I have found a way to remote desktop with the 970 plugged in and gotten GPU-Z to show some info. Most notably, not much can be retrieved about the card. In MSI Afterburner no OC options or temps are available.