Hello folks,
 
First of all, thanks for the help and the effort done on this forum, it's really cool to see :)
 
I have built a new gaming rig recently. Here's the specs :
- Ryzen 3700x, cooled by an Noctua NH-14S
- Asus Strix B450i Gaming
- G.Skill 16 Go DDR4-3600
- Corsair SF600
- Ncase m1
 
All in all, pretty classic Ryzen gaming build for the Ncase, which is a small factor case. Built went smoothly, Windows started without a problem.
 
I have also bought an Artic Accelero 3, in order to cool down my new shiny (and beefy) graphic card. I have built all my gaming rig over the last years, so I felt confident to be able to install correctly the Accelero 3 and benefit some really nice temperature.
 
So I went on and installed the Accelero 3 on the graphic card. Again, no problems to report, built went smoothly, and I was able to install the Accelero 3 and put 2 Noctua NF-F12 below it. As I don't have currently the adapter to handle the fans directly by the graphic card, I just plugged the two Noctua directly to the motherboard, and set the fan to max speed (which is quite noisy, but I am ready to accept that until the adapter come).
 
And the results were quite good :
Idle temp :
 - CPU : around 43°C
 - GPU : around 36°C
 
Load temp :
 - CPU : I touched 73°C under Prime95, after one hour
 - GPU : In game, with the Accelero, I got 65°C after half an hour of "The Forest" gaming, all parameters set to Max.
 
So I know that it's not a real benchmark per se, but my internet broke at home for more than one week and I was not able to download graphic benchmark software.
 
Now, here's the issue : yesterday, I played The Forest half an hour roughly, with OSD for CPU and GPU temp, and everything was butter smooth. 120 FPS, graphic card was running full speed but I'm positive that the GPU temp stayed stable around 65°C, which was impressive.
 
And suddenly, the whole PC just shut electrically. I tried to restart the PC and nothing. I removed the power cable, waited for a couple of minutes, then re-plugged it and restarted it, and the PC booted fine.
 
So, I started to debug the PC.
CPU is probably fine, considering the one-hour of benchmark I put to it.
I ran one night of Memtest on my RAM, 4 pass without any error.
GPU is probably not suffering from heat, judging by the current temperature I've got.
I have also reinstalled Nvidia drivers (446.14) completely.
 
So I'm posting this to try to have more insight on how I could try to debug this problem, especially if the problem is on the graphic card (which is not sure yet).
 
Could it be VRM / Mem overheat ? Sadly, my graphic card doesn't have temp sensors for those parts.
 
I also hop that it's not the graphic card which is faulty, because to install the Accelero 3, you need to remove the original cooler. I have read that it's not a warranty void, but I'm fearing that the shop where I bought the graphic card could think it's a warranty void and not reimburse me the card. Considering the price, I would be be more than disappointed.
 
Thanks for the help !
 
Sterfield
 

Hmm that's a good point.
 
Thanks for the link, I have checked it and added all my stuff in it. Outcome for a Ryzen 3700x + MB + 16 GB DDR4 + 2080 Super (with the correct GPU clock + memory) + 1 SSD ~= 500 W, which was my target.
 
I'll try to scavenge a more beefy PSU and test with it.
 
In the meantime, and as I hadn't got the possibility to run a proper GPU benchmark (and also because my internet access was finally fixed), I bought a 3Dmark licence and ran some benchmark. Time Spy in normal version and intensive version both passed without a problem. GPU temp was around 65-70° the whole time.
 
Then I ran Time Spy 3 time in stress test (20 loops each, around 15 min) then 2 times Time Spy intensive (20 min each) and all went well, no crash / sudden shutdown.
 
I'm puzzled…

This PSU is brand new.
 
I'll try to find how much power the PC is pulling.
 
Thanks for the advice.

Do you have your PC plugged into a UPS?  I had a similar issue recently.  It ends up my PC was exceeding my UPS system's power limit.

I had simmilar issue and found out that
a) my 3.3v was low (like "*(removed by Cool GTX) it is 3.0v!") and
b) my 12v was low under cpu load (like went down from 12.3 to 11.8), while under gpu load it was fine.
c) this problem was getting worse and worse, like from shutdown once a week to couple shutdowns per day.
 
 
I tried to work with 24pin contacts (cleaning, pushing to get better contacts), that gave me a minor improvement. And so I ended up with ordering new cables (still waiting) and removing all cables from my system, checking them (all looks fine), checking connectors on both psu and mother board (fine too) and reconnecting it all without cable management (because I still wating new red-and-black power cords :) ).
Routing cables without any hard bends (I had especially tight circle bend near psu and I believe that this elastic tension gave me bad contact on my 24pin  from psu side) gave me great improvement: now my system doesn't shut down at all again, 12v drops down under cpu load to 12.2, but 3.3v still a bit low, like 3.28v.
 
But there is more: suddenly I discovered that my USB type c on front pannel didn't work and even more: connecting something to it made my bios turn off PC due to USB over current after POST. It turned out that you actually could connect that ugly USB 3.1 gen2 wire upside down and that would short-circuit something to the ground. And that is awesome: despite that these header and cable have a key that should prevent it from connecting upside down - that actually don't work and you can do it easy :)
After I fixed this problem my 3.3v had went up to normal 3.32v, but overal system stability was achieved before with good contact on 24pin motherboard cable.
 
I can guess that these shutdowns was motherboard "fault" or some anti-surge protection, because:
a) I tried to keep my system shutdown if power was lost (there is option in bios called like "Restore on AC Power Loss") and also I know that any psu protection just shut it down, but not power cycle your system;
b) I tried to disable any OS restarts on any faults - still power cycled;
c) I actually repeated that issue with touching and "shaking" 24pin connector and that had gave me power cycle even if my motherboard was instructed to power off after AC power loss.
But sad thing that there was no messages after post while I hoped to saw something like this:

 
PS this maybe ugly english grammar of mine is result of reading "english grammar for dummies", so I am kind of trying to use something exept present and past simple :) Please be forgiven, it will be better I hope :)
 
 
 edited your post Cool GTX ... language removed

That was 3 letters, not 4, if you ask me :)
But ok, next time I will use another 3 letters "OMG" :)