Possibly dying PSU unit?

[dDule]

Currently running PSU: CoolerMaster MWE 650W V2 80+ Gold Fully Modular

 

Hey, i've been dealing with PC crashes without complete shutdowns and just blackscreens with no control that force me to manually shut it down on power button.

 

I've been running an R9 270X, then R9 290, R9 290X, 2x RX 580s and now an MSI GamingX Duo RTX 3060Ti and almost all have been crashing with the same issue, which kind of assured me it's not the GPU as it's not possible that ALL GPUs crashed the exact same way...Random blackscreen while gaming( nothing while browsing the web or watching videos), most of the time the sound buzzes, and theres no control over the system anymore. PC doesn't shut down, so i have to force shutdown while holding the power button.
 

Before i concluded it's my GPU once again, i wanted to investigate a little bit more, and i think i've found the culprit.

 

 

While on desktop, with HWInfo turned on with 4 live graphs and NO load, voltage on GPU is dropping down to 11.485v.

 

The situation is that the RTX 3060Ti CANNOT run any big stress tests immediately. It will instantly result in a crash about 0.5-1 second later after the test starts. The only one that it will pass is the 3D Adaptive which lets me start from 5% load and go up to 100% in increments. That one passes without any issues all day long. Even when i run this test and it doesn't crash for about an hour, i tried running the 100% load 3D stress test, and it passes it sometimes, but only if i stress the GPU for a while and immediately start running the heavy test. Power test always fails 1 second later... Any other 100% tests fail immediately and i have even tried to start at 5% but same issue. Perhaps the type of load is different even if it's the same amount of %load...

VRAM test passes any day.

Anyway...in a sense, "Cold" starts fail easily. The 1st picture shows the drop in voltage right after the test start and the bottom one shows the voltage at 100% on the GPU.

These voltages are coming from sensors on the GPU itself, i have tested the PSU with multimeter on itself with a pin short but doesn't show changes in voltage and is pretty flat on all 8pins and the 24pin(It's within tolerance ~ 11.9). I haven't tested the rails with a multimeter while stress testing due certain physical constraints, but i will try that soon, aswell.

Can anyone conclude that this is a PSU issue or is there any other chance the GPU is at fault even though last 5 AMD gpus were crashing the same way? I have used a GT210 for some time and i have it for troubleshooting purpouses and it never crashed, but it also doesn't use PSU cables, only pulls power from the PCIe slot. I did have this same PSU while using all previous GPUs and i have it for around close to 5 years. Are there any other possible tests i can do that will help me in resolving this? Thanks!

[thekingofmonks]

What does event viewer say?

 

Any other parts you can switch to test with?

[freeagent]

Time for a new, stronger PSU. 11.2-11.4 is when mine would dump too.

 

RIP