1 Frame frozen on monitors crash while system under load

[Ikaunieksplay]

Hello

Recently I have been experiencing issues with my PC where it "crashes" while under load. The crash just freezes a frame on my monitors, no sound, but the RGB still lights up, fans spin and such. This doesnt happen in one specific program, just when under load. It crashed loads of time when trying to render a decently complicated video in Premiere Pro and while rendering a photo of a lego model in bricklink studio.

PC Specs:
CPU: Ryzen 5 2600

CPU Cooler: Stock Cooler
Mobo: ASUS Prime B450M-K

GPU: PowerColor Red Dragon RX580 8Gb
PSU: Thermaltake TR2 S 500W
SSD: Intenso 128gb
HDD: Seagate Barracuda 1tb
Case: Fractal Design Focus G

I know that my power supply isnt exactly high quality, but it doesnt seem to be the culprit here to me because like I said earlier the lights are still on and the fans spin. Any clue what this could be?

[Albert F]

I am still a bit skeptical about that PSU. Have you tried reinstalling your gpu drivers and clean installing windows?

[Ikaunieksplay]

11 minutes ago, Albert F said:

I am still a bit skeptical about that PSU. Have you tried reinstalling your gpu drivers and clean installing windows?

I have reinstalled the GPU drivers, but nothing changed. Havent clean installed windows, didnt think that could help. You think I should do it?

[Albert F]

3 minutes ago, Ikaunieksplay said:

I have reinstalled the GPU drivers, but nothing changed. Havent clean installed windows, didnt think that could help. You think I should do it?

you have nothing to lose other than some time

[oscar7601]

I suppose you have not overclocked anything in the system?

[Ikaunieksplay]

6 minutes ago, oscar7601 said:

I suppose you have not overclocked anything in the system?

Nope, everything is at stock.

[oscar7601]

Is your computer reasonably clean inside? No huge piles of dust?

What you describe looks like instability to me. This usually should not occur in a stock configuration except if there are serious thermal issues.

[Mark Kaine]

Well this sort of stuff is precisely why people advise against getting low quality power supplies... sure i can't say for certain,  but especially the under load thing implies the psu is simply giving up and you should probably replace it before it takes other components with it. 

[oscar7601]

3 minutes ago, Mark Kaine said:

Well this sort of stuff is precisely why people advise against getting low quality power supplies... sure i can't say for certain,  but especially the under load thing implies the psu is simply giving up and you should probably replace it before it takes other components with it. 

 

It very well could come from a bad PSU but I think we should rule out all of the other possibilities before ever talking someone into throwing (potentially good) hardware in the bin.

 

Plus, a stock 2600 and an RX580 really are not demanding components to power.

[Ikaunieksplay]

8 minutes ago, oscar7601 said:

Is your computer reasonably clean inside? No huge piles of dust?

What you describe looks like instability to me. This usually should not occur in a stock configuration except if there are serious thermal issues.

It is clean, I have cleaned it quite recently. I have had thermal issues a while back, but then it fully crashed and after rebooting showed me that the CPU overheated. Don't know if this is particularly relevant in this case.
 

 

6 minutes ago, Mark Kaine said:

Well this sort of stuff is precisely why people advise against getting low quality power supplies... sure i can't say for certain,  but especially the under load thing implies the psu is simply giving up and you should probably replace it before it takes other components with it. 

I would like to know for sure that it is the PSU before replacing anything(just like oscar said), since I am quite tight on money right now and dont want to waste any.

[oscar7601]

I'd be tempted to tell you to try to underclock lower the power going to the GPU, just to see if the same components are able to pass the same "stress test" (doing the same rendering with the same settings in Adobe Premiere) with only the GPU clock (and thus power hunger) lowered.

 

This can be very easily done and undone. You just need a software like MSI Afterburner (FYI it doesn't matter if your GPU is not of the MSI brand  ). Set the power limit waaay down and try to run your demanding render again.

 

If the same computer successfully passes the test then it would probably narrow the issue down to either being the PSU or maaaaybe (but realistically unlikely) a GPU requiring more voltage to keep stable under heavy load, in the case of quite heavy sillicon degradation (after being used for mining with non-conservative settings for example, but this is really unlikely IMHO)

 

FYI : Setting the power limit down will effectively diminish the power set available for the GPU, which will in turn lower its running speed accordingly. So the render may take a bit longer to run.

[Ikaunieksplay]

5 minutes ago, oscar7601 said:

I'd be tempted to tell you to try to underclock lower the power going to the GPU, just to see if the same components are able to pass the same "stress test" (doing the same rendering with the same settings in Adobe Premiere) with only the GPU clock (and thus power hunger) lowered.

 

This can be very easily done and undone. You just need a software like MSI Afterburner (FYI it doesn't matter if your GPU is not of the MSI brand  ). Set the power limit waaay down and try to run your demanding render again.

 

If the same computer successfully passes the test then it would probably narrow the issue down to either being the PSU or maaaaybe (but realistically unlikely) a GPU requiring more voltage to keep stable under heavy load, in the case of quite heavy sillicon degradation (after being used for mining with non-conservative settings for example, but this is really unlikely IMHO)

 

FYI : Setting the power limit down will effectively diminish the power set available for the GPU, which will in turn lower its running speed accordingly. So the render may take a bit longer to run.

Guess I will try that. Don't think it could be the degradation thing because I havent mined on this card at all, and in general don't really play super demanding games that push it a lot minecraft ftw
 

Another thing that might help is that the video always crashed during the render at exactly the same spot, 12.1% if I remember correctly. I only managed to render it by switching the output format to ProRes instead of H265. Don't know if this tells you anything.

Will come back in a few hours with the results

The odd thing that totally slipped out of my mind is that the video render crashed even when I turned off GPU acceleration, once again at the same percentage. Although oddly enough when I turned off GPU acceleration in Premiere the render didn't slow down even a little bit. I dont know exactly what this means but just nice to know.

Doesnt change the fact that it crashed during other loads too, along with the PC just feeling "slower", and other apps crashing randomly..(For example yesterday opened a new tab in firefox, that crashed firefox AND discord for some reason. Just seems like the system is less stable overall.)

[Mark Kaine]

7 hours ago, Ikaunieksplay said:

I would like to know for sure that it is the PSU before replacing anything(just like oscar said), since I am quite tight on money right now and dont want to waste any.

Still, ideally you could try out a better psu to eliminate that possibility.  The other way around (testing other "stuff") is a lot more tedious and might end up being non conclusive. 

 

I'm just telling you that it sounds like pretty typical psu issue,  doesn't mean it ends up being that,  but it's where I'd probably start.  

 

And yes, actually undervolting your gpu,  would be another way trying to figure it out. 

 

So idk set it to 80% power limit or something and lower clocks and test it for stability. 

 

 

[Ikaunieksplay]

UPDATE 1:

Couldn't really get to testing over the weekend. Finally got some time this evening,  I decided to do a stock test of the same video project and it crashed again, same spot. I restart the pc using the disk reset button as always, and it comes up with the error message saying that the CPU overheated.

 

I find this very odd, because never did it show the overheating message when it crashed before, and the temps seemed ok to me. 

 

Guess I was wrong, I cleaned the system and ran out of time, so hope to do the test tomorrow.