I built my computer a little under 2 months ago, for the first few weeks everything was fine but now I have been dealing with crashes of increasing frequency. When it crashes it just instantly shuts off and reboots, no stalling no blue screen, just an instant reboot, no dump files are created and event viewer only shows a Event ID 41 Kernal-Power Critical Error "The system has rebooted without cleanly shutting down first." with a bugcheck code of 0. Also occasionally it would fail to boot up, stalling during "Installation of PCH Runtime Services" and the motherboard debug led would light up for VGA error: GPU not detected or failed, but would boot correctly after turning off and back on. It started crashing maybe one out of every 10 times I launched Mass Effect Andromeda, always crashing at roughly the same spot during the loading screen. Then it started crashing every time it loaded. Then other games (Sniper Elite 4, Dying Light, Destiny 2) started crashing, also during loading screens. I updated drivers, re-seated all hardware, still crashed. Flashed motherboard BIOS, clean installed windows, still crashed. Now it started crashing during gaming, not just on loading screens. But crashes are not predictable, and sometimes it will go for several hours without issue. Running monitoring software shows that most temperatures are well under their limits, with GPU 1 and 2 sensors rarely going above 60. The only outliers are the Mem2 and Mem3 modules, which are routinely much hotter than the rest of the card (for instance when Mem1 is at 48C, Mem2 and Mem3 are at 66C and 68C respectively, and that is after setting a much more aggressive fan curve in Precision X1, 30% at 30C 60% at 50C and 100% at 70C on all 3 fans, was even higher with default curve), and my CPU never even really goes above 50C. I have run Firestrike stress testing and gotten 98.7% stability, stays totally stable during Prime95 torture testing, ran four passes of MemTest86 with no errors, chkdsk with no errors, UserBenchmark shows every component working above expectations, and I ran pretty much everything else I can think of and found no issues. Voltages seem pretty stable, but (and I'm not sure that this matters) as soon as I am under load I get a PerfCap reason of VRel and sometimes Pwr, VRel. Unfortunately I have neither an extra GPU to swap it out for, nor another computer I can swap this GPU into, so I am limited to testing with my current equipment. Nothing in my system is overclocked (except for the XMP for the RAM, but disabling XMP did not stop the crashing) even though the Scan feature of Precision X1 says I could set it to +80. I am pretty much at a loss for what to do, the only hint I have gotten is the VGA error during PCH Runtime Installation which points to the GPU. Do I have a faulty GPU? Is there something else I am missing? What do I do next? Any help would be appreciated. If there is any information I forgot to supply, or any tests I should run and post results of, I would be happy to.
 
These are my system specs
Case: CoolerMaster h500m Mesh
CPU: Intel i7 9700K
CPU Cooler: EVGA CLC 280
GPU: EVGA 2080 ti FTW3 Ultraa
RAM: 16GB (2x8GB) G.Skill TridentZ RGB 3200 cl16
HD: 1TB Samsung 970 EVO Plus and 500GB Samsung 860 EVO 
PSU: Corsair AX860 860w 80+ Platinum
Motherboard: MSI MEG Z390 ACE
 
EDIT: Included motherboard on system specs

Yes and no. The PSU is certified factory refurbished, but was purchased at the same time as the other components (March) and only used in this computer. Could the PSU cause the VGA failure on boot? And are there any signs I could look for in the voltages that would indicate a PSU issue? Voltages seem stable (for instance +12V rail stays between 12.000V and 12.096V), is there any way to test the PSU without just buying a second one and swapping it in?

The psu or video card could cause this.

Pretty much, but you could use a digital multimeter to see if the 12v rail drops out of spec when the crashes occur. That would point to a faulty psu. Are you powering the card with two separate pci-e power cables? If no, do that first as it could be the problem.

860w is enough for your system.

No problem. No your mem temps were fine. The memory can be fine up to 95c.

Just one thing to point out with the fan curves. When you goto set up a custom curve. Remember that your setting the temp curve to that component. So GPU fan from the GPU temp. The mem from one of those gpu chips (no idea which its sensing from.. some chips run much hotter then others) the VRM for the power fan.

So use the ICX button to check those temps.. set the curve accordingly.

I like to get the fans moving max on load so they can do their job. So 80% at 40c for the GPU and 100% at 45% and up.. so this would be more effective if the other 2 fans would work the same since that heatsink is all connected to the GPU die.

Would be best if we could sync it all to the gpu temp.. but I dont think that works. So one needs to watch what the load temps are for the memory and power areas to make sure the fans ramp up as well to help cool the GPU temps.

If you have a 2nd monitor.. have PX1 open on that monitor with ICX open to watch those temps while gaming or benchmarking and then adjust the fan profile accordingly.

Basically have the fans running at like 30-50% at idle to keep the temps in check before it ramps up to help the overall load temps.

Dont let everything cook and then expect to be able to control the temps on air once it's already really higher over 50-60c at idle.