Question Abrupt PC power outages (overcurrent protection?)

Bunnyhunter

Commendable
Sep 27, 2020
5
0
1,510
Hi,

first my system setup:
CPU: Ryzen 7 5900x
Board: Asus Crosshair viii Hero wifi / MSI Tomahawk
GPU: STRIX GTX3090 on a PCIe 4.0 Riser
PSU: 1200Watt BeQuiet Dark Pro 12
RAM: 32GB G.Skill Trident Z RGB für AMD DDR4-3200 DIMM CL14 Quad Kit
SSD: M.2 NVME Samsung Pro 250GB OS, +1TB Samsung Pro additional space
The entire thing is custom hardpiped and powered by EK-Quantum Kinetic TBE 300 D5 PWM D-RGB - Acetal,
thermals are not an issue according to hardware monitor
Display is on 1920x1080

Now to the issue:
My system keeps abruptly blackscreening under heavy load. Event viewer code is 41 - 63 kernel power error,
which leads me to believe it is
most likely due to the PSU safety features kicking in.
I am not 100% sure yet if GPU or CPU load is the cause. Possibly both. The system has been running for
just about 3 years with rare black/blue screen restarts.
I assume i never really put it through its paces enough for it to reach the limit.
Now that system requirements for games have increased i might just have reached it.
The issue can easily be reproduced when i try to deploy in Tarkov or run Heaven Benchmark or
start the MSI afterburner oc scanner.

What i have done so far: Not kidding i literally put my RAM sticks into the fridge
since the ASUS board was known to have RAM issues..... it actually got me 20 seconds
further into the deploy on Tarkov 😀 - other than that to result
Removed RAM sticks for single and dual, no result
Replaced Motherboard from ASUS to MSI, no result.
Set PCIe back to 3.0 no result
Disabled Precision Boost for CPU -> kinda feels like it improved it
Limited GPU Frequenzy to 1500/1600 starting at 800mV -> enabled me to at least deploy in Tarkov
If i stand on the edge of the Tarkov map and look in - it tends to crash.
If i get into an engagement - it tends to crash.

Maybe someone here has a better guess than me on what might be happening.
My feeling is the power consumption spike of the 3090 is strong enough to
tear the whole system down? Then again its a 1200 Watt PSU and a (supposedly)
good quality one as well. From what i know Tarkov is a CPU heavy game so it could
also be the CPU, Heaven on the other hand is clearly GPU focused.
There have been reports on transient power consumption spikes for the 30-Series
cards which might cause blackscreens, maybe this is the issue here...

My next step is to replace the PSU or even have GPU and CPU/Motherboard
run on 2 different PSUs. I dont think the issue is CPU/GPU health related or caused by bad
memory as i have checked all sticks.

If anyone has had similar experiences with 3090 black screen i would love
to hear other suggestions i can try before i order another 300$ PSU :/

Love and Happiness

Bunny
 
Try plugging gpu directly into motherboard. Reason: Risers have been hit/miss.
Yes, I know; it would be a pain with a hard loop, but you gotta do what you gotta do.

OCP tends to be followed by fans ramping up.
 
At times pressing the start button on the pc shows no response. In that case i have to fully cut power to the system for it to come alive again.

Fans ramp up and then failure occurs.
 
At times pressing the start button on the pc shows no response. In that case i have to fully cut power to the system for it to come alive again.

Fans ramp up and then failure occurs.
So sometimes you need to unplug or cut the PSU off and back on for it to do anything and sometimes you don't? I mean if you don't do that and push the power button nothing at all happens?
Your using 4X8GB of memory was this bought in one package of 4 sticks or 2 packs of 2?
 
RAM was bought as a quad stack 4x8 GB.

So sometimes you need to unplug or cut the PSU off and back on for it to do anything and sometimes you don't? I mean if you don't do that and push the power button nothing at all happens?

That is correct. Sometimes it blacks out and i can press the buttons all i want, unless
i disconnect power fully and in that probably give the PSU a full reset nothing happens.
Then there are the black outs where i can just go back to the normal power button of
the pc and restart.
And then there are those instances where the system will automatically restart by itself
like 2-3 seconds after the blackout happend.

Try plugging gpu directly into motherboard. Reason: Risers have been hit/miss.
The GPU was plugged in directly when i initially build and tested everything air cooled.
The riser has been there for 3 years now as well and since i already used two different risers (PCIe 3 and PCIe 4) to test if there could be an issue i dont really think thats where the problem lies.

Thank you very much for the feedback so far!!!!

Bunny
 
Hey, just to complete this thread - i bought a MSI Ai1300P ATX 3.0 PSU and the issues are gone. It seems my old PSU got bad over time. Thank you for ur help!

Bunny
 
It seems my old PSU got bad over time.
They 'age' over time, so what the unit can output now vs when you got it year one, is less.
They can age faster depending on the paces you put them through.
Also, if one uses cable extensions, they may have to replace units more frequently; the extra length leads to greater rail voltage droop.