Question Kernel power error 41 on new pc : (

[Fayerss]

Hello ,bought this pc 1 month ago for my father to play some games with him. First week and it was having kernel power error 41(62) while playing star citizen,(error was causing a reset every time), brought it back to seller and they changed motherboard and psu but told us that they werent able to reproduce the problem .

Now after 1 to 2 hours of playing any intensive game(we tried elden ring and star citizen) we still get the same error... : (.
i deactivated the reset if critical error option to try to get more info and see what happens, after 1 hour pc started artifacting very bad and we were forced to manually shutdown.


Those are the specs :
Intel Core i5-12400F
Motherboard : Asus PRIME H610M-A D4
Inno3d GeForce RTX 3060
2x 16 Gb TeamGroup 3200Mhz DDR4
ATX Xigmatek X-Power 600W / 80PLUS

Temps are perfectly fine, pc runs again after each reset : /
Im kinda at a loss atm to how to pinpoint the problem,prime95 test run without problem for 12 hours...., furmark gpu stress test with together with 12 thread cpu burn worked fine for 2 hours, only gaming seems to cause it...
I just bought occt and trying to run some tests to try to reproduce the problem(3d adaptive in 5% increase with 2 min steps un to 90%) so we can avoid to take the pc back again if it doesent pass them (store is 70km away....)

sadly this shop doesent seem very competent
any guidance on how to pinpoint/reproduce the problem would be greatly appreciated

 

[drivinfast247]

I would start buy replacing that PSU with a quality 650w unit. Look to see if the Corsair RM/RMx series is available in your area.

 

[DSzymborski]

Yeah, that PSU is absolutely frightening to pair with an RTX 3060; I wouldn't have even turned on this PC with that connected.

Since the PSU should have been replaced under every and all circumstances -- it's a very cheaply made, group-regulated unit -- replacing the PSU with something that should ever have been used with this system would be the obvious place to start.

 

[Fayerss]

i mean , its a prebuilt and i just bought it so the shop will have to replace the parts , im trying to pinpoint the problem since they dont seem very competent , pc is running trough all benchmarks and stability test just fine : / , my objective here is to pinpoint the problem or reproduce it in order to dont get the same crap again

 

[Fayerss]

it just went trough occtp test for 3 hours , now im testing ram for another 3 hours

 

[Phaaze88]

Sorry, but can you be a little more specific on what some of these benchmarks and stability tests were?
Some games have their in-game benchmarks that are actually LESS demanding than the game is, so the best stress test is the games themselves.
Other gpu benchmarks are more synthetic and don't actually represent games either, being less stressful than the games.
OCCP has multiple tests that can be done. What was tested for 3 hours?

 

[Fayerss]

Sorry, but can you be a little more specific on what some of these benchmarks and stability tests were?
Some games have their in-game benchmarks that are actually LESS demanding than the game is, so the best stress test is the games themselves.
Other gpu benchmarks are more synthetic and don't actually represent games either, being less stressful than the games.
OCCP has multiple tests that can be done. What was tested for 3 hours?

so :, crash occurs with : star citizen (ranging from 30 min to 2 hours),
elden ring (half hour last time)

benchmarks :
no crashes with 12 hours prime 95
no crashes with furmark : native resolution, max antialising and cpu burn runned at same time for 3-4 hours
no crashes with 2 hours occp psu test(alimentation)
no crashes with occp gpu adaptive(the pay version one) test 3 hours, 5% increase , 2min timestep up to 90% usage
no crashes with occp gpu free test , 3 hours
no crashes with occp memory free test , 3 hours
no crashes with occp memory stability payd version test, 3 hours


now pc is running haven benchmark 4.0 free version with:
directx 11 ,
quality: ultra,
tasselation: extreme
stereo 3d: off
multi monitor: off
anti alising 8x
native resolution

gonna add occp psu test on top of that if it fails to crash for 2-3 hours , then retry crash time with elden ring

 

[Phaaze88]

Gaming gpu loads are harder on power supplies than cpus are.
Prime 95 is cpu, cpu cache, memory, and a very stead load throughout. Constantly changing loads are the kicker. This app isn't it for ruling out psu issues.
Furmark is also steady, and doesn't change load as much as the actual games can.
Just checked out OCCP's psu test... wow, that's not a realistic load at all...
OCCP memory isn't going to do it.
I don't have the paid version, so I can't do the gpu adaptive that you did.
Heaven Benchmark is dated.


Try Display Driver Uninstaller and install a different one.

Try Msi Afterburner: unlink the power and temperature limits(the chain link icon), and experiment with lower power limits.
95%. Click apply, then test. Crashes?
90%. Repeat. Crashes?
85%.

 

[Fayerss]

ok gonna try that now thx , what exactly should i uninstall and install with the display driver uninstaller? , got last drivers for videocard from geforce exerience , installed it anew when the pc came back from shop the first time

 

[Phaaze88]

ok gonna try that now thx , what exactly should i uninstall and install with the display driver uninstaller? , got last drivers for videocard from geforce exerience , installed it anew when the pc came back from shop the first time

Uninstall the current driver, and try the 2nd most recent WHQL driver package. During the driver install, choose the option to not install Geforce Experience; maybe - just maybe - the settings it uses for ER and SC are too aggressive. It's supposed to optimize performance, but it's not perfect.

 

[Fayerss]

runs fine with 85% power limit should i run a benchmark during this? which one? im running haven atm

 

[Phaaze88]

That's your power supply's limit, if that's the case.

 

[Fayerss]

That's your power supply's limit, if that's the case.

i mean , its running fine up to 85% should i go down even more? , it doesent seem to do anything

 

[Phaaze88]

i mean , its running fine up to 85% should i go down even more? , it doesent seem to do anything

No, there's no need to go further if it stops crashing at 85%.

 

[Fayerss]

No, there's no need to go further if it stops crashing at 85%.

ah i see so this should fix it and not reproduce the crash , no idea if it crashes as no benchmark managed to crash it yet ...trying to run elden ring now for a few hours with this settings,lets hope it works
wouldnt this affect performance of the gpu?

 

[Phaaze88]

Yes and no, all because of the Gpu Boost algorithm built into the card.
It automatically and dynamically overclocks the core clock beyond the base depending on operating temperature and how frequently the power limit is hit. There are several boost curves depending on those parameters.
Lowering the power limit can result in a negligible decrease in performance, because operating temperature doesn't get as high.

 

[Fayerss]

Yes and no, all because of the Gpu Boost algorithm built into the card.
It automatically and dynamically overclocks the core clock beyond the base depending on operating temperature and how frequently the power limit is hit. There are several boost curves depending on those parameters.
Lowering the power limit can result in a negligible decrease in performance, because operating temperature doesn't get as high.

crashed after 30 min elden ring with 85% : (

 

[Phaaze88]

Well, try going all the way down on the power limit... :/
There's not much further to go, I figure, since TPU's list of Inno3D 3060 vbioses all have a maximum low limit of 59%.
If it still crashes on that... I may not have a choice but to agree with the other 2 posts about the power supply. It is pretty low on the ladder here(Tier E: Avoid).

The only other thing I can think of that would be giving you trouble in ER and SC and not other apps, is some driver(not gpu) or software overlap causing errors, and that's going to be a pain to narrow down without trying a clean Windows install.

 

[zygarte]

If you can confirm that it isn't a hardware or psu issue, then you should try to disable gpu or vga driver, usually, a random kernel power without blue screen is related to a needed driver, gpu, chipset ... if that so, windows wouldn't fix it because the most of needed hardware can be used (in a limited way) without factory drivers.

 

[Hybenos]

I have a similar setup and having the exact same issue. I have tried it all! (replied in another thread)

Reinstalling windows several times
Memcheck with 8 passes (no errors)
Disk checks
Disconnected everything and put it all back one piece at a time (disks, ram etc).
Removed mobo to check for faults
Tried all the giudes with energy settings and so on

I even bought a new PSU to try and it still fails.

My guess is maybe the AMD processor?

I have unchecked the box where windows reboots when critical errors occurs. I do not like this at all but it seems to be the only way.


Mobo: ASUS B550F-gaming
CPU: Ryzen 7 5800x
RAM: Corsair Vengeance Pro (2x16GB) 3600Mhz
GPU: ASUS ROG RTX 3080 10gb
PSU: Corsair RM, 850W Gold

 

[Technerd18]

have you taken into account is may be your house or apartment on old wiring that mat not be able to transfer a certain amount