Recently upgraded to a 3080 TI FTW3
CPU: Ryzen 5900x
Mobo: MSI MPG Gaming Edge b550
PSU: EVGA 850W G2 80+ gold (using separate 3 VGA/PCIE connectors on the PSU to power the GPU)
 
Under load, Wi-Fi starts to fail then the PC will BSOD or just black screen and crash. I thought that 850W would be enough based on EVGA's power meter recommendations. After some research I've found that this is likely due to transient power spikes common in the RTX 30 cards. I've tried setting power limits to as low as 50% on the card with no increase in stability. I'm just posting to see if there's something I could be missing before I go ahead and get a new PSU.
 
Update:
I tried a Seasonic Focus GX 1000w PSU and I still ran into the same issues. I realized after some tests and research that the GPU crashes due to voltage spikes (Or at least the crashes coincide with these spikes). I can get it relatively stable by setting the voltage curve and limiting it to .925V at 1750Mhz. It's stable but an unacceptable performance penalty. It seems like anything above .95 is not very stable. (This is the same with both PSUs). I think there is an issue with the card, but I'll try a few more things including disabling the wifi from the BIOS to see if that does anything. I could have resolved this sooner if I had another PC to test. I'm going to a friends place later this week to test out the GPU in his machine. Thanks for the advice everyone! 
Thanks for the advice everyone!
 
PS: I am not using cables from different power supplies. I've seen a couple people advising me against it. Thanks for the warning, but I'm not sure what I said to make people think I was doing this. I'm definitely not : )
 
Edit: I think I was able to figure out the issue. I started getting these motherboard USB overcurrent protection warnings when booting up my PC. It would happen every once in a while. I looked around the USB ports on the case and motherboard as well as the headers on the motherboard for damage that may be causing a short. After enough frustration I eventually decided to test everything outside of the case. I attached everything outside of the case and used the original PSU and all of a sudden, no crashes! I realized it had to be something with the case (or the RGB from the AIO since I left it unplugged for this test). I checked to make sure the standoffs were the correct positions. And carefully placed the motherboard back into the case, but only plugged in the power switch/reset/led pins. The PC still seems stable with the GPU at stock settings. Over the course of the next few days I'm going to test the case USB as well as the RGB to see what it is causing that's causing the issue. I'll have an update soon. Fingers crossed! Very happy I don't have to RMA GPU.

Thanks. I wouldn't say I skimped out on the PSU. I continued to use what was otherwise a good quality PSU that's still under warranty and above the recommended power output. Plenty of people have had no trouble with 850W and in some cases less with similar specs. I'm just posting to see if there is more I can do to verify that the issue is the PSU not being able to  accommodate the GPU and not that the GPU or PSU is just faulty or possibly some other issue.

Absolutely NEVER use other brand replacement cables for most PSUs, use only the  ones that came with your specific PSU or replacement ones for your specific PSU from the manufacturer. Not even cables that come from the same company for a different model unless recommended by the manufacturer. In fact most PSU companies will tell you that in the manual.
Don't believe me?  

My 3080ti runs fine with the power limit maxed / and oc'd on a corsair rm750 (750 gold).  Not saying it isn't your PSU, but I am saying the wattage should be able to handle it.

you should not mix modular cables with different powers supply's if you do bad things will happen.

slaander
 b550 is not top of line board for running 5900x and 3080

Are you saying the board isn't able to supply enough power or that the pcie lanes is being saturated or something? I don't think that's the case unless the board is defective. There are plenty of b550 builds with 3080/3090s. I'm not saying it can't be an issue with the board but I imagine the issue would be a defective board not that the board isn't high end enough. Luckily a friend is letting me borrow their PSU to test.

Ryzen 5900x, 3080ti ftw3, B550 Gaming carbon with Evga G3 750w PSU
runs fine for me, however, i dont use the wifi onboard and disabled them from the bios.
I kept the 750w because I have had it and it's been reliable. (will upgrade my psu when price are better :x)
 
I thought the reason ftw3 require 3rd 8pin is to help ease the power draw.
 
maybe try unplug every cable including the 24pin, and reconnect those, make sure they are properly seated.
can also try update driver for the onboard intel wifi.

andyliumd
Ryzen 5900x, 3080ti ftw3, B550 Gaming carbon with Evga G3 750w PSU
runs fine for me, however, i dont use the wifi onboard and disabled them from the bios.
I kept the 750w because I have had it and it's been reliable. (will upgrade my psu when price are better :x)
 
I thought the reason ftw3 require 3rd 8pin is to help ease the power draw.
 
maybe try unplug every cable including the 24pin, and reconnect those, make sure they are properly seated.
can also try update driver for the onboard intel wifi.

I tried a Seasonic Focus GX 1000w PSU and I still ran into the same issues. I realized after some tests and research that the GPU crashes due to voltage spikes (Or at least the crashes coincide with these spikes). I can get it relatively stable by setting the voltage curve and limiting it to .925V at 1750Mhz. It's stable but an unacceptable performance penalty. It seems like anything above .95 is not very stable. (This is the same with both PSUs). I think there is an issue with the card, but I'll try a few more things including disabling the wifi from the BIOS to see if that does anything. I could have resolved this sooner if I had another PC to test. I'm going to a friends place later this week to test out the GPU in his machine.

This makes me nervous to run my 3080ti ftw3 with the 750w gold that came in the bundle, but maybe I'll just wait until there's a hiccup to buy a new psu

Just read your edit.  USB was something I had trouble with when I first built my machine as well.  Ended up being something with my keyboard I was testing with during the build phase of my computer.  Most of the time I wouldn't even get to a load screen (black screen).  Weird thing is that the keyboard works just fine on the machine now.  No telling what I was running into during my build...
 

Mine turns the pc completely of under gaming load

with the bundle...   that wasn't perchance a "newegg" bundle was it?  Make sure you didn't end up with the leftovers and its not one of the models that have issues.

Tcosby5t
Straight from EVGA

If its completely shutting off during load, that sounds like CPU overheating.  Find you some software to check your temps at idle and while playing a game.  With all the safeties in place, if the CPU reaches a certain point the whole thing will shut itself off. Make sure your cpu fan is working, and in a instance of overheating you may have to repaste the cooler or blow out any dust that may be built-up on the cpu cooler.
 
Its either that or your using a Power supply that can't handle the load. 
 
 

Tcosby5t
I run a new 850w evga gold psu.
My ftw3 3080 runs just fine it's when I put my ftw3 3080ti in that it turns off.

I'm guessing a 3080 vs 3080 ti driver is the same driver.  But it is a different card.  still could be heat of the card itself.  run something like gpu-z and check what the temps get at while gaming.
 
Only other thing I can think of is underclocking a bit and see if it stabiles.  I'm pretty sure the power requirements didn't change that much between the 3080 and 3080ti.
 
If its not overheating, and underclocking doesn't stabilize it, its time for RMA

Tcosby5t
I run a new 850w evga gold psu.
My ftw3 3080 runs just fine it's when I put my ftw3 3080ti in that it turns off.

I've got a 1300w Seasonic Prime Gold that was totally fine on a 3080 FTW3 at 450w sustained, but would go into protect shut off on a 3080 Ti FTW3.  IMHO they pull power differently.