RTX 3090 crashing

[Omer6543]

Hey,

My pc crashed today after like 20m of playing a game and then the motherboard showed a red light and the gpu light was off and the fan was working,

I checked with the motherboard website and it showed that is has a problem with the GPU I tried to reset the pc and it showed the red light and no input

I waited a few minutes and it worked so I did a stress test and everything was fine after 10 minutes in the stress test it crashed and on the second try the system froze 

and on windows reliability I saw those error codes : 141 - 117

 

and now the pc is working fine I did another stress test and it passed it

 How can I 100% know its the gpu?

 

my setup :

amd ryzen 3700x

16gb ram 3200hmz

1tb nvme

PSU antec hcg gold 80+ 750W

and rtx 3090FE

b450 Auras elite

 

[Coaxialgamer]

The way to know for sure is to test things with another GPU.

That being said, 750W does seem a bit tight for a 3090 build, especially considering the power spiking tendencies of the 3090. It's possible that a combination of temps and changing benchmark conditions are pushing it over the edge after a little while.

[Omer6543]

8 minutes ago, Coaxialgamer said:

The way to know for sure is to test things with another GPU.

That being said, 750W does seem a bit tight for a 3090 build, especially considering the power spiking tendencies of the 3090. It's possible that a combination of temps and changing benchmark conditions are pushing it over the edge after a little while.

Thank you, I don't have another gpu

and I know its a bit tight but I checked and the system takes around 650 -700w max so its fine 

and I have it more then a year with a lot of stress unity, blender etc do I don't think its from the power spiking 

I think its the temp I saw the memory jump to 102 and its crashing only when there is a stress on the gpu

like now the windows and everything works only crash when stress test or game

and yesterday it was working fine  nothing happened to damage it or something the last thing I remember that I got blue screen but nothing else

[Coaxialgamer]

6 minutes ago, Omer6543 said:

Thank you, I don't have another gpu

and I know its a bit tight but I checked and the system takes around 650 -700w max so its fine 

and I have it more then a year with a lot of stress unity, blender etc do I don't think its from the power spiking 

I think its the temp I saw the memory jump to 102 and its crashing only when there is a stress on the gpu

like now the windows and everything works only crash when stress test or game

and yesterday it was working fine  nothing happened to damage it or something the last thing I remember that I got blue screen but nothing else

Something like an HD5450 or GT210 from ebay could do the trick if you're so inclined, though I understand if you aren't.

 

I'm less concerned about the steady-state power consumption of the rig (which is how Nvidia gives out PSU recommendations after all), but the 3000 series as a whole tends exhibit spikes of greatly increased power consumption when first loaded, which can trip some PSUs up. That said, if you haven't had issues for over a year it probably isn't the problem.

 

The fact that things are back to normal are a tad puzzling admittedly. Perhaps try reseating cales and other connectors around the GPU? Otherwise, if everything is back to normal I wouldn't worry too much about it given there's not a whole lot you can do to isolate the problem to begin with.

[GuiltySpark_]

Frankly it sounds like the GPU is failing. That follows a pretty standard failure mode. Black screen, fans still going. If you reboot the PC manually it works again. 

 

If the PSU was tripping a protection circuit it would have shut off completely. I ran a 3090/5800x on a 750W PSU for two years just fine, a tiny little SFX one even. It is a very high quality PSU though so the power rating alone doesn't tell the whole story. 

 

Can you confirm the PC appears to still be on when this happens (minus the screen going black) and doesn't fully shut down? 

 

 

[Omer6543]

11 minutes ago, GuiltySpark_ said:

Frankly it sounds like the GPU is failing. That follows a pretty standard failure mode. Black screen, fans still going. If you reboot the PC manually it works again. 

 

If the PSU was tripping a protection circuit it would have shut off completely. I ran a 3090/5800x on a 750W PSU for two years just fine, a tiny little SFX one even. It is a very high quality PSU though so the power rating alone doesn't tell the whole story. 

 

Can you confirm the PC appears to still be on when this happens (minus the screen going black) and doesn't fully shut down? 

 

 

Thank you, the pc is on the fans and everything are working and the RGB is on only the gpu lights was off and the fan on

but the modo showed a red light on the VGA and he didn't let me boot so I shut it off and after 5 minutes its working fine

 

19 minutes ago, Coaxialgamer said:

Something like an HD5450 or GT210 from ebay could do the trick if you're so inclined, though I understand if you aren't.

 

I'm less concerned about the steady-state power consumption of the rig (which is how Nvidia gives out PSU recommendations after all), but the 3000 series as a whole tends exhibit spikes of greatly increased power consumption when first loaded, which can trip some PSUs up. That said, if you haven't had issues for over a year it probably isn't the problem.

 

The fact that things are back to normal are a tad puzzling admittedly. Perhaps try reseating cales and other connectors around the GPU? Otherwise, if everything is back to normal I wouldn't worry too much about it given there's not a whole lot you can do to isolate the problem to begin with.

Thank you, what do you mean reseting cales ?

its working fine now but only when I am not in game 

If i enter a game or do 2 -3 stress test he will crash 

 

[Mark Kaine]

3 hours ago, Omer6543 said:

Thank you, I don't have another gpu

and I know its a bit tight but I checked and the system takes around 650 -700w max so its fine 

unless you have a way to measure transient power spikes that's quite the optimistic assumption  ...

 

 

[CommanderAlex]

10 minutes ago, Mark Kaine said:

unless you have a way to measure transient power spikes that's quite the optimistic assumption  ...

 

 

And no Kill-A-Watt meter will ever catch that as they aren't that sensitive.

 

The computer black screens and restarts, getting stuck on a VGA LED?? Does indeed sound like a graphics card problem and needs RMA. Happened to my 6900XT and it was indeed the card defective. 

[Mark Kaine]

9 minutes ago, CommanderAlex said:

And no Kill-A-Watt meter will ever catch that as they aren't that sensitive.

yeah, i know,  but obviously there's a way to measure it... prob need some "pro equipment" ¯\_(ツ)_/¯ 

 

as for, uh, the issue on hand, 700w usage + 100 to 200w "spikes" could easily trigger some protections on less than ideal power supplies,  just something that shouldn't be dismissed (and also this can happen under no load too, just certain spikes can be enough to trigger the protection as psu will see this and think "uh, oh, better shut it down!" )

[CommanderAlex]

4 minutes ago, Mark Kaine said:

yeah, i know,  but obviously there's a way to measure it... prob need some "pro equipment" ¯\_(ツ)_/¯ 

Mhm, yes indeed. 

 

4 minutes ago, Mark Kaine said:

as for, uh, the issue on hand, 700w usage + 100 to 200w "spikes" could easily trigger some protections on less than ideal power supplies,  just something that shouldn't be dismissed (and also this can happen under no load too, just certain spikes can be enough to trigger the protection as psu will see this and think "uh, oh, better shut it down!" )

Yeah there are other variables to consider and some questions to be answered by the OP. This power supply is a good unit basing from the PSU Tier List so it should be able to handle the transient spikes. 

 

3 hours ago, Omer6543 said:

Thank you, what do you mean reseting cales ?

I believe he meant "reseating cables", meaning to unplug and fully plug in the power cables again. 

[GuiltySpark_]

Said it before but this is almost certainly GPU failure. Power delivery system probably getting too hot running a benchmark/game and a defect or even a cracked solder joint being affected and the GPU shuts down. In these cases the rest of the system stays on. Some have reported they can even still hear the game or whatever playing in the background.

 

 

 

[Omer6543]

14 hours ago, GuiltySpark_ said:

Said it before but this is almost certainly GPU failure. Power delivery system probably getting too hot running a benchmark/game and a defect or even a cracked solder joint being affected and the GPU shuts down. In these cases the rest of the system stays on. Some have reported they can even still hear the game or whatever playing in the background.

 

 

I retested the gpu today

 and  after 5 tests the pc froze and  I hear the game playing in the background and on the second time the pc just crashed and when I turn it on its just the red led with the fans on and gpu lights are off

but without stress its working fine

thank you everyone for helping me