Gpu usage spikes to 100% and crashes, even when idle

[purifiedprune]

All of this started when I was playing Fallout 4. I tried to bring up the pip-boy and the game crashed. At first I thought this was a problem with the game itself,but it started happening everywhere including Windows when I bring up the taskbar.

My first thought was to reset my overclocks in MSI Afterburner which lessened the frequency of the crashes but they still happened.I tried rolling back drivers,resetting bios settings,tried debug mode but it still happens every now and then.Eventually I lowered the core and memory clock down by 100 which seemed to fix the problem but it only feels like a temporary solution.My overclocks were at +175 core and +500 memory.

My specs:

CPU:i5-6600k

GPU:Asus ROG Strix GTX1070(Driver 436.48)

Ram:Corsair Vengeance LED 16GB(8GBx2)

Motherboard:Asus Z170 Pro Gaming Aura

PSU:Corsair RM650x(2016)

[Jelly Donut]

You could have a bad GPU.

Perhaps you can try gaming on your CPU graphics/I Graphics and see if the issue occurs?

If it doesn't crash, consider RMAing your 1070.

[Zireth]

If you have another GPU you know works, put that in and see if the errpr occurs then.

If it does its another component

If it doesn't, your GPU or drivers may be faulty.

[SeriousCCIE]

Do you get any on-screen "artifacts" or glitches?

It could be overheating.

You don't mention the temperature of anything, just the core utilization. How hot does it get when active, and how hot does it stay when idle?

[purifiedprune]

Origineel geplaatst door SeriousCCIE:

Do you get any on-screen "artifacts" or glitches?

It could be overheating.

You don't mention the temperature of anything, just the core utilization. How hot does it get when active, and how hot does it stay when idle?

Sorry,forgot to mention the temps. The card stays around 68-70 degrees celsius when gaming and 39-41 when idle. I don't think overheating is the issue here though.

As for artifacts or graphical glitches,no. I have not encountered anything that resembles both in the time I have used this PC.

[Jelly Donut]

Origineel geplaatst door mr groovy man:

Origineel geplaatst door SeriousCCIE:

Do you get any on-screen "artifacts" or glitches?

It could be overheating.

You don't mention the temperature of anything, just the core utilization. How hot does it get when active, and how hot does it stay when idle?

Sorry,forgot to mention the temps. The card stays around 68-70 degrees celsius when gaming and 39-41 when idle. I don't think overheating is the issue here though.

As for artifacts or graphical glitches,no. I have not encountered anything that resembles both in the time I have used this PC.

Temps are fine. As I mentioned, try playing on i Graphics and see if your system crashes.

[Snow]

Just the other day I was playing around with a card that had some weird screws holding the cooler and it ran fine for few seconds, but then all of a sudden it showed temps at around 100 degrees, video driver crashed, and then it ran as lowest clocks it can, resulting in 100% GPU usage and low framerates. It appeared those screws that were used instead of original ones didn't press the cooler hard enough, so at some part of graphics chip there was a small gap, and when that exact part was used - thing did just overheat. It's unlikely to be the save with your card, but absolutely do keep an eye on temps just before the crash happens, as it might be just cooler mounting problem.
Meanwhile, what you said about lowering clocks by 100MHz solving it - it sounds like this exact card fails to maintain factory overclock, and while it's technically okay to see OC failing - it'snot you who OCed it but ASUS, so hold them responsible for that and ask for a refund or another card of the very same model. Or try going with Inno3D/EVGA, those are exclusive NVidia brands I personally have trust in.

[purifiedprune]

Origineel geplaatst door Hot Paprika Soup:

Origineel geplaatst door mr groovy man:

Sorry,forgot to mention the temps. The card stays around 68-70 degrees celsius when gaming and 39-41 when idle. I don't think overheating is the issue here though.

As for artifacts or graphical glitches,no. I have not encountered anything that resembles both in the time I have used this PC.

Temps are fine. As I mentioned, try playing on i Graphics and see if your system crashes.

Okay,I'm using the IGPU right now and so far so good, no crashes and games seem to be working fine.

[purifiedprune]

Origineel geplaatst door Snow:

Just the other day I was playing around with a card that had some weird screws holding the cooler and it ran fine for few seconds, but then all of a sudden it showed temps at around 100 degrees, video driver crashed, and then it ran as lowest clocks it can, resulting in 100% GPU usage and low framerates. It appeared those screws that were used instead of original ones didn't press the cooler hard enough, so at some part of graphics chip there was a small gap, and when that exact part was used - thing did just overheat. It's unlikely to be the save with your card, but absolutely do keep an eye on temps just before the crash happens, as it might be just cooler mounting problem.
Meanwhile, what you said about lowering clocks by 100MHz solving it - it sounds like this exact card fails to maintain factory overclock, and while it's technically okay to see OC failing - it'snot you who OCed it but ASUS, so hold them responsible for that and ask for a refund or another card of the very same model. Or try going with Inno3D/EVGA, those are exclusive NVidia brands I personally have trust in.

Thanks for the input.
About what you said about factory overclock,my card was working just fine earlier this week at +175 core clock and +500 memory clock so imo I don't think the factory oc is failing,just the card itself.

[Snow]

Origineel geplaatst door mr groovy man:

Origineel geplaatst door Snow:

Just the other day I was playing around with a card that had some weird screws holding the cooler and it ran fine for few seconds, but then all of a sudden it showed temps at around 100 degrees, video driver crashed, and then it ran as lowest clocks it can, resulting in 100% GPU usage and low framerates. It appeared those screws that were used instead of original ones didn't press the cooler hard enough, so at some part of graphics chip there was a small gap, and when that exact part was used - thing did just overheat. It's unlikely to be the save with your card, but absolutely do keep an eye on temps just before the crash happens, as it might be just cooler mounting problem.
Meanwhile, what you said about lowering clocks by 100MHz solving it - it sounds like this exact card fails to maintain factory overclock, and while it's technically okay to see OC failing - it'snot you who OCed it but ASUS, so hold them responsible for that and ask for a refund or another card of the very same model. Or try going with Inno3D/EVGA, those are exclusive NVidia brands I personally have trust in.

Thanks for the input.
About what you said about factory overclock,my card was working just fine earlier this week at +175 core clock and +500 memory clock so imo I don't think the factory oc is failing,just the card itself.

Wait, so it crashed with factory OC or with yours? Or both?

Tho either way failing is failing, give it back.

[purifiedprune]

Origineel geplaatst door Snow:

Origineel geplaatst door mr groovy man:

Thanks for the input.
About what you said about factory overclock,my card was working just fine earlier this week at +175 core clock and +500 memory clock so imo I don't think the factory oc is failing,just the card itself.

Wait, so it crashed with factory OC or with yours? Or both?

Tho either way failing is failing, give it back.

Yeah,it crashed with both,sorry for not including more info in the OP.
I guess I have to send it back.