PC cuts power mid game

Leprechaun312

Leprechaun312

Leprechaun312

Soldato
Joined
1 Jun 2012
Posts
2,817
Location
UK
Hi, about to yeet my pc through a window. I've included the spec below but basically im having an issue where mid game or mid running hashcat my pc will just power cycle. No event logs. I swapped my 3080 for a 6900xt and it's something to do with this.

At a bit of a loss as to what to do, computer had absolutely no problems with the 3080 in the same games and with hashcat so I'm not convinced the PSU is at fault. When this first started happening if i gave the card a nudge the problems would go away for a few days, which makes me suspect the card is so ******* heavy it might be coming out of the PCI slot, or the PCI slot is faulty. I have a riser cable and the case supports vertical mounted GPUs however I'd need to order an AIO for it to fit....

Just wanted a sanity check before i order an AIO or PSU could the PCI slot being faulty / not being perfect in the slot be the issue? I don't even think the card is faulty, as there are no other issues outside the power cut. I've done this as well from another forum post having similar issues to see if it solves it.

3900x
32gb kingston hyper x (can find the exact model if needed)
ASUS ws-x570 ace motherboard
xfx merc 6900xt
corsair hx1000i PSU
 
"Modern" high end graphics cards are indeed insanely heavy.
So that heavy weight and then thermal expansion from all the heat GPU dumps out during continuous load could cause issue with the slot.
I would suggest looking for trying to support the card to see if that fixes it for long term.
 
Little update, been away for a few days, left pc on as i needed access to it while i was away. Came back and it was off, turns out it bluescreened.
These are pretty unspecific but from googling the ntkrnlmp.exe one is usually GPU / GPU driver issue.... faulty card? I can't see drivers causing PC to reset.
Ordered a new GPU out of frustration with this, will DDU and swap tomorrow see if anything changes, looking like an RMA jobbie tho

ZKdEwlU.png
 
Leprechaun312 said:
Little update, been away for a few days, left pc on as i needed access to it while i was away. Came back and it was off, turns out it bluescreened.
These are pretty unspecific but from googling the ntkrnlmp.exe one is usually GPU / GPU driver issue.... faulty card? I can't see drivers causing PC to reset.
Ordered a new GPU out of frustration with this, will DDU and swap tomorrow see if anything changes, looking like an RMA jobbie tho

ZKdEwlU.png
Click to expand...
I've had that error before when overclocking the VRAM too much.
 
Are you overclocking anything? If so, turn those overclocks off, go back to standard and see if the issue goes away.

Try a GPU support bracket, or failing that, just lay your entire PC on it's side for a few days so the GPU sits upright.

What are the various temperatures at when it power cycles?

The pictures you posted suggests to me the problem is the system RAM. Do you have any spare old RAM you can swap in temporarily?
 
Ormy1 said:
Are you overclocking anything? If so, turn those overclocks off, go back to standard and see if the issue goes away.

Try a GPU support bracket, or failing that, just lay your entire PC on it's side for a few days so the GPU sits upright.

What are the various temperatures at when it power cycles?

The pictures you posted suggests to me the problem is the system RAM. Do you have any spare old RAM you can swap in temporarily?
Click to expand...

No overclock, as for the support bracket i i found some adjustable stand thing and used that system had been fine for about a week, but now the bluescreen has happened.
Temp wise, depends on what i was doing maybe 70 in game, closer to 85 in hashcat, all well below thermal shutdown

Dont have any spare DDR lying about, i mean i can do memtest tonight see what happens
 
Another update, swapped in the 3080, no power losses mid game and over two hours running hashcat with the same load that caused issues with 6900xt, same deal no power losses.
Started the RMA process, think the card is just faulty.
 
That still doesn't mean that card is deffo faulty (although it is likely), there are other possibilities like corrupted drivers or corrupted registry entries or something like that. Only way to check is either put the 6900XT in another machine or do a format and reinstall of windows and all drivers.
 
So update. Had one blue screen and one power cut since swapping out gpu. Ram memtest and seems one of the sticks is faulty. Have identified which and running a memtest on the other 3 to check it's just the one. Confirmed as well that it's not the slot on the motherboard 100% the stick. I suspect my gpu RMA will be rejected :cry::cry:
 
This will probably not be helpful

I suggest taking a hard look at CPU/memory controller.
Bad memory will normally manifest as program crashes or blue screens. Random restart is something a CPU would trigger.

And it could perhaps be remedied by adjusting SoC voltages.
 
alec said:
This will probably not be helpful

I suggest taking a hard look at CPU/memory controller.
Bad memory will normally manifest as program crashes or blue screens. Random restart is something a CPU would trigger.

And it could perhaps be remedied by adjusting SoC voltages.
Click to expand...

So i do agree, it's the weirdest manifestion of ram issues i've seen. Would expect just loads and loads of BSODs or the pc to legit refuse to boot, not random power cuts
The third screenie tho, its done 4 passes with the two sticks, so im guessing the IMC is fine? i would expect errors on all sticks.
 
alec said:
I was going to reply IMC has to work twice as hard with 4 sticks.
But if it reliably errors out on one stick, and then works with 2, its probably the RAM indeed.
Click to expand...
two sticks, errors galore 20 secs into mem check.

The other two zero errors for 4 passes ~3 hours

Tests were done in the same slots, and then individually with the faulty sticks just to confirm.
 
You must log in or register to reply here.
Back
Top Bottom