6700XT black screen crash then automatically restart, WHEA logger 18 Cache Hierarchy Error & Kernel-Power 41

[jupiter_man]

Hi all,

I get a black screen crash while I play battlefield 5 / God of War / Assassins Creed Valhala, the black screen will last for around 5 seconds then the computer will automatically restart.
Most of the time it happens randomly, but it keep happening during the last boss fight + cut scene in God of War so I am stuck in that loop.
It could crash any time during the actual fight or the cut scene after that, totally random.
And it never happen in FIFA22.

Then when I check Event Viewer I found:
WHEA-Logger 18, Cache Hierarchy Error, Processor APIC ID: 1 to 13 (yes after so many crashes I have got crash report on all 13 of them)
Kernel-Power 41, task 63.
I used DDU to unstall and reinstall both GPU and sound card driver.
updated the mother bios to the latest non beta version
No overclock on cpu and gpu
disable XRP
Have tried both Ryzen 3600 and Ryzen 5600X
Both cpu and gpu temp are around 70c.
when it crash the RGB light on my motherboard, graphic card and case are all on

After so many crashes I have swaped my 6700XT to my old R9 Fury, then it didnt crash during the boss fight + cut scene, tested it for 5 times and no crash.

Below is my spec:
Ryzen 3600 / 5600X
MSI B450 TOMAHAWK MAX
Sapphire NITRO+ AMD Radeon RX 6700 XT 12GB GDDR6 Graphics Card
Kingston Fury Beast 16GB (2x8GB) DDR4 PC4-25600C16 3200MHz Dual Channel Kit
EVGA SuperNOVA 750 G2, 80+ GOLD 750W
Crucial MX500 CT500MX500SSD1 500 GB

Thanks for reading!


I totally dont know what to do so any help would be much appreciated! THANKS!

 

[jupiter_man]

How old is your PSU?

Its only two years old, and all the RBG lights are working normally, so if it is not working properly (like voltage spike) then maybe it should affect those light as well?

That whea error is for the CPU not GPU

Go to your bios and load default profile, don't use XMP, don't PBO or any overclocking. Then check if it still crashes

GPU failure is more likely to produce visual artifacting and if the game crashes it produces graphics driver failure message, whea 18 is for CPU crashing/losing power

Yes thats what I thought initally so I changed my CPU from 3600 to 5600X, both are brand new when I open the box.
I have disabled the PRO then had the crash within like 10 seconds into the boss fight, when PRO is on AUTO it normally last longer before crashing...
And I had no visual artifacting before this crash or on any other game.
Will it be the motherboard?

Keep Ripple disabled

I was going to add to @Grim5 observation that when I seen the temps if they are going north of 70 it could be airflow/heat/VRM. I would also while troubleshooting put GPU fan in a fixed range like 50-75%, I never trust that zero fan thing.

tried different fan function and none work as well...
And I had higher temp after hours of gaming and it didnt crash, so I think it will not be the temperature problem on both cpu / gpu.

Download prime95 and run small fft test on all cores. This will test you’re cpu under very heavy load I’d run it for 20 minutes. Use hwinfo to monitor temps. If you have a weakness with you cpu it will find it. If this ok. Down load fur mark. Test for 20 mins this will test your gpu. Report back you results.

Just run it for 2 hours and passed 1680k test on all 6 cores and didnt crash.
Temp around 70c

Will test furmark later and report back.

Kernel power 41 can be related to insufficient power. I was getting exactly the same black screen followed by a hard reboot and Kernel power 41 when I switched to a RTX3070 from my old GTX1070 with a high quality 650w Gold rated psu. I put it down to the power spikes that RTX3000 cards suffer from so bought a 2021 Corsair RM850x and the problems were gone. I wouldn't have expected a 6700Xt to have the same problem, especially with your high quality 750w psu but I suppose there is always a chance. Are you using two seperate pci-e cables for the gpu or one cable with the daisy chained pair of connectors? If using one cable try switching to two seperate cables and see if that makes a difference. Another thing to try is undervolt the gpu. This can actually add performance while reducing power draw which is the opposite to what you would expect to happen when reducing voltage. I have done it on my 3070 and it cut power draw by up to 120w (game dependant) while having the card boosting higher.

I am using two pci-e cables, will try to switch them and report.
I did try to lower the voltage on AMD software from default 1200mV to 1170mV or 1150mV, from max frequency 2714MHz to 2424Mhz, still crash...

 

[jupiter_man]

Its only two years old, and all the RBG lights are working normally, so if it is not working properly (like voltage spike) then maybe it should affect those light as well?





Yes thats what I thought initally so I changed my CPU from 3600 to 5600X, both are brand new when I open the box.
I have disabled the PRO then had the crash within like 10 seconds into the boss fight, when PRO is on AUTO it normally last longer before crashing...
And I had no visual artifacting before this crash or on any other game.
Will it be the motherboard?




tried different fan function and none work as well...
And I had higher temp after hours of gaming and it didnt crash, so I think it will not be the temperature problem on both cpu / gpu.






Just run it for 2 hours and passed 1680k test on all 6 cores and didnt crash.
Temp around 70c

Will test furmark later and report back.




I am using two pci-e cables, will try to switch them and report.
I did try to lower the voltage on AMD software from default 1200mV to 1170mV or 1150mV, from max frequency 2714MHz to 2424Mhz, still crash...

Just ran furmark for half an hour and didnt crash

 

[jupiter_man]

I had the same sort of issues. Bought another 850W, Corsair rm850x, kernel 41 issues gone. It replaced a 7 year old hx850 so was due for a change.

Thanks will try to get another one and report back.
Just replaced old cable with new cables for GPU and CPU, insert them in other slots, still no luck.

Did you try to change the motherboard?