Is this a GPU problem? Screen gets corrupt and pc restarts

[Sincercow]

When I'm playing a game some time randomly the screen get black and looks corrupt, the only thing that is visible its bright stuff/white text (its hard to explain, see images). there is no blue screen just shows this corrupted screen then restarts after like 10 seconds. This was not happen frequently (about a every week to 2 weeks) but recently (playing elder ring) It crashed about 10 minutes in. I get a kernel-power 41 error (every time) . I got a new graphics card in January and the crashes started about a month ago. Temp of gpu and cpu are fine/not overheating. I have checked the ram with programs and nothing is wrong with it. I believe it to be the graphics card and wondering if it could possibly get fixed so I wouldn't have to do a RMA. Also, Should I try another gpu to see if its the issue or is this definitely a gpu problem?

 

some of my specs:

Amd Ryzen 7 5800x (have it undervolted but its probably not causing the problem)

Ram: 2x16GB of 3600MHz (g.skill)

Gpu: ASUS ROG Strix NVIDIA GeForce RTX 3080 V2 OC Edition (on P mode) 

850 wat psu 

Mother board: ROG CROSSHAIR VIII FORMULA

 

Pics have information I dont want to share blacked out.

[Deadpool2onBlu-Ray]

7 minutes ago, Sincercow said:

When I'm playing a game some time randomly the screen get black and looks corrupt, the only thing that is visible its bright stuff/white text (its hard to explain, see images). there is no blue screen just shows this corrupted screen then restarts after like 10 seconds. This was not happen frequently (about a every week to 2 weeks) but recently (playing elder ring) It crashed about 10 minutes in. I get a kernel-power 41 error (every time) . I got a new graphics card in January and the crashes started about a month ago. Temp of gpu and cpu are fine/not overheating. I have checked the ram with programs and nothing is wrong with it. I believe it to be the graphics card and wondering if it could possibly get fixed so I wouldn't have to do a RMA. Also, Should I try another gpu to see if its the issue or is this definitely a gpu problem?

 

some of my specs:

Amd Ryzen 7 5800x (have it undervolted but its probably not causing the problem)

Ram: 2x16GB of 3600MHz (g.skill)

Gpu: ASUS ROG Strix NVIDIA GeForce RTX 3080 V2 OC Edition (on P mode) 

850 wat psu 

Mother board: ROG CROSSHAIR VIII FORMULA

 

Pics have information I dont want to share blacked out.

Are you using three separate 8 pin cables straight from your psu? Or are you using a splitter? 

[Sincercow]

8 minutes ago, Ryan829 said:

Are you using three separate 8 pin cables straight from your psu? Or are you using a splitter? 

all separate 

[Deadpool2onBlu-Ray]

1 minute ago, Sincercow said:

all separate 

Try turning off p mode. Leave everything stock and see if it crashes 

[Sincercow]

1 hour ago, Ryan829 said:

Try turning off p mode. Leave everything stock and see if it crashes 

turned it off seems to not be crashing for now. It seems to be using the same amount of watts. how would Q mode be causing the problem?

[Sincercow]

ooops

 

[Sincercow]

1 hour ago, Ryan829 said:

Try turning off p mode. Leave everything stock and see if it crashes 

It crashed but with no corruption just went black and restarted.

Still a kernel-power41 error too 

[Deadpool2onBlu-Ray]

1 hour ago, Sincercow said:

It crashed but with no corruption just went black and restarted.

Still a kernel-power41 error too 

RMA

[Sincercow]

On 3/31/2022 at 6:36 PM, Ryan829 said:

RMA

I did a furmark gpu stress test 2 times, one which lasted probably about 30 minutes then crashed, I was streaming in discord and had a 4k video running (Kernel-power41 error too). The other one lasted 1 hour and 30 minutes which just had monitoring software open. I stopped it because it was going too long It didn't crash. 

I noticed the ram to be 94-96C hot which is weird for it being only at 10% usage for both tests, for the first test my case fans were at there lowest and for most of the second test they were at there highest. This seems like the GPU not working but also something could be overheating which I have no clue on what that could be. Also the TDP goes above 100% of that means anything. 
 

Ill probably start the RMA prosses the weekend though.

Here are some pictures of it running the test (test 2): 

[Robchil]

set it down to 100% power? 

 

[HoosierTransfer]

kernel power 41 means that the pc rebooted unexpectedly so i think it is a gpu problem

[lafrente]

Those mush images are generally the result of high temps. 

[Sincercow]

I seemed to have fixed/Made the crashing happen less (haven't had enough time to conclude anything) I followed some other threds and it seemed to be a ram issue in the clock speed, voltages, or the FCLK frequency. For me it was the FCLK frequency It was on auto so I changed it to halve of my RAM speed (1800MHz). I have gotten a crash on a bench mark but it actually blue screened and other tests it didn't crash so seems to have somewhat fixed it. Ill mark this as the solution of I get no crashes for a while/It it doesn't corrupt my screen again. 

 

I have done about a 5-6 hour run of elder ring with no crashes and about and 1-2 hour of streaming it on discord with no crashes(usually crashes wile doing this) so everything SEEMS stable. 

[Sincercow]

ok it crashed again so i guess it didn't work

I guess RMA time

 

[Sincercow]

When i debut with windbg it says the problem is AuthenticAMD. Is there something wrong with amd software or the cpu itself, the BIOS?

[Sincercow]

Whelp, seemed to be the undervolted or the curve optimizer. I had it to all cores, negative, with the magnitude at 10, I moved it to 0 and after more than a month no crashes. I may try the magnitude at 5 but for now I'm keeping it at 0. I'm pretty sure this was the problem, only time will tell.