Gaming Random Restarts

[Johnsenish]

So i have a issue related to the PC restarting when i play games (never happens in idle)
When i play WoW on my CRG9 it tends to restart about every 15-20mins, but when playing Apex on it, its behaving nicely. Same goes for playing Tarkov on my HUA271.

I've checked temps (sitting at 50-60 degrees C*) when gaming.
The windows log only gives me generic info about Kernel power, unexpected power loss.

Just to make sure, ive cleaned my parts nicely, even taken it out of the cabinet with no help.
A quick google search points me toward the PSU or GPU..
I have gotten a accepted warranty return for my GPU and PSU, but i can't afford to be without my PC for a month or three in these times (Work with IT from home) 

At a loss what else to do.

Any tips or tricks before i have to send it in?

Will resetting a CMOS really help me out here?

 

 

• Windows 10 - 64 bit
• Intel i7 7700k 
• 2x8GB Corsair Vengeance RAM 3200mhz (XMP profile on)
• RTX MSI Gaming Trio X 2080 Ti
• EVGA Supernova G3 (850W 80+ gold)

[PineyCreek]

Try a stress test like Furmark to see if it restarts maybe.

[RONOTHAN##]

It really does sound like a bad PSU. The way I find to determine if it is, go into MSI Afterburner and lower the power limit to something like 60 or 70%, maybe even lower and see if it still crashes. If that fixes it, it is definitely either the GPU or PSU. To determine which, raise the power limit back up, and drop the card's clock speed by 100MHz. if it doesn't crash, it's 90% the GPU. if that doesn't fix it, it's the PSU. You can also experiment with lowering the GPU memory clock, but that would usually cause artifacting and game crashes, not system crashes

[Johnsenish]

1 minute ago, RONOTHAN## said:

It really does sound like a bad PSU. The way I find to determine if it is, go into MSI Afterburner and lower the power limit to something like 60 or 70%, maybe even lower and see if it still crashes. If that fixes it, it is definitely either the GPU or PSU. To determine which, raise the power limit back up, and drop the card's clock speed by 100MHz. if it doesn't crash, it's 90% the GPU. if that doesn't fix it, it's the PSU. You can also experiment with lowering the GPU memory clock, but that would usually cause artifacting and game crashes, not system crashes

I'll give it a go once i come back from work.

 

9 minutes ago, PineyCreek said:

Try a stress test like Furmark to see if it restarts maybe.

Dunno if there is a difference, but i've tried that Heaven stresstest for 2 hours without any issues.
Also done that memtest86 without any error reported.

[RONOTHAN##]

8 minutes ago, Johnsenish said:

Dunno if there is a difference, but i've tried that Heaven stresstest for 2 hours without any issues.
Also done that memtest86 without any error reported.

Heaven isn't that intensive of a stess test, so that only tells you that your GPU isn't super dead. Is there a specific scenario where you can get it to crash all the time, like a certain scene in a game?

[Johnsenish]

23 minutes ago, RONOTHAN## said:

Heaven isn't that intensive of a stess test, so that only tells you that your GPU isn't super dead. Is there a specific scenario where you can get it to crash all the time, like a certain scene in a game?

Playing WoW, but only when running dungeons or raids, then it happens about every 15min. AFK WoW in a city and its all fine.
its so illogical to me i don't get it.

 

THO! im gonna be completely honest.. could it be my thermal paste on CPU? or the AIO being bad?

[RONOTHAN##]

17 hours ago, Johnsenish said:

Playing WoW, but only when running dungeons or raids, then it happens about every 15min. AFK WoW in a city and its all fine.
its so illogical to me i don't get it.

 

THO! im gonna be completely honest.. could it be my thermal paste on CPU? or the AIO being bad?

It is possible, check your CPU temps to confirm. If you're going up into the 80s and 90s with an AIO, that's your issue

[Johnsenish]

On 1/7/2021 at 9:53 AM, RONOTHAN## said:

It really does sound like a bad PSU. The way I find to determine if it is, go into MSI Afterburner and lower the power limit to something like 60 or 70%, maybe even lower and see if it still crashes. If that fixes it, it is definitely either the GPU or PSU. To determine which, raise the power limit back up, and drop the card's clock speed by 100MHz. if it doesn't crash, it's 90% the GPU. if that doesn't fix it, it's the PSU. You can also experiment with lowering the GPU memory clock, but that would usually cause artifacting and game crashes, not system crashes

So after having the PSU at 60% for 24 hrs, there is no crash.

So i am giving the 100mhz negative clock speed a try now, should i have it go lower then that if its stable?
Still i dont understand how if the clockspeed is 100mhz less its a sign of GPU being the fault?

[RONOTHAN##]

1 hour ago, Johnsenish said:

So after having the PSU at 60% for 24 hrs, there is no crash.

So i am giving the 100mhz negative clock speed a try now, should i have it go lower then that if its stable?
Still i dont understand how if the clockspeed is 100mhz less its a sign of GPU being the fault?

So the reason why I have you do both is that when you lower the power limit, the card draws less power, but it also lowers its clockspeed, and a lower clockspeed usually means a more stable card. By raising the power limit back to normal, it isolates the GPU, making sure that the GPU draws full power, but still doesn't clock up all the way, so if the GPU is unstable at its factory clockspeeds, it will usually be stable again at the lower clockspeed. 

[Johnsenish]

46 minutes ago, RONOTHAN## said:

So the reason why I have you do both is that when you lower the power limit, the card draws less power, but it also lowers its clockspeed, and a lower clockspeed usually means a more stable card. By raising the power limit back to normal, it isolates the GPU, making sure that the GPU draws full power, but still doesn't clock up all the way, so if the GPU is unstable at its factory clockspeeds, it will usually be stable again at the lower clockspeed. 

So by now i would normally have atleast 2-3 restarts the last 2hours, since nothing is happening, u would say that the GPU is the culprit here? 


Considering there is 4-5 months waiting for a new RTX 30 series card here, will the 100 mhz matter alot in games? not so keen on doing a warranty with the long deliver times (maybe in 6-7 months its better, and a warranty return would be better)

[RONOTHAN##]

1 minute ago, Johnsenish said:

So by now i would normally have atleast 2-3 restarts the last 2hours, since nothing is happening, u would say that the GPU is the culprit here? 


Considering there is 4-5 months waiting for a new RTX 30 series card here, will the 100 mhz matter alot in games? not so keen on doing a warranty with the long deliver times (maybe in 6-7 months its better, and a warranty return would be better)

100MHz in games won't matter too much, maybe 5-10 fps. I would still contact MSI about returning the card and see what the RMA time would be. If it's less than a month, I would just do it.

[Johnsenish]

1 minute ago, RONOTHAN## said:

100MHz in games won't matter too much, maybe 5-10 fps. I would still contact MSI about returning the card and see what the RMA time would be. If it's less than a month, I would just do it.

Got a feeling i jinxed myself, 2 restarts in last 5 minutes, been doing nothing  different then last 3 hrs..

[RONOTHAN##]

1 minute ago, Johnsenish said:

Got a feeling i jinxed myself, 2 restarts in last 5 minutes, been doing nothing  different then last 3 hrs..

So It's probably the PSU then. EVGA does simultaneous RMAs, so you send it to them while they send it to you (at least the last I remember). If you're outside of the warranty period, just order a new one and leave the GPU on reduced power limit until the new one gets there.

[Johnsenish]

18 minutes ago, RONOTHAN## said:

So It's probably the PSU then. EVGA does simultaneous RMAs, so you send it to them while they send it to you (at least the last I remember). If you're outside of the warranty period, just order a new one and leave the GPU on reduced power limit until the new one gets there.

I'll do that! Thanks for your help, much appreciated. 
i will mark the thread solved