Question CPU overheating only during full screen games ~65-70

[internalmemory]

This started to happen about a month, month and a half ago. I think I first noticed it in Forza Horizon 4, which had me move to different games. Then I noticed it in Splinter Cell Blacklist, and at the time it was the only other game that would make my CPU spike instantly to PC complete-freezing/crashing temps. That game (at the time) could turn my CPU temps from 30 to 75 in a matter of 30 seconds regardless of the DirectX or any other graphical settings.

But now every single game does this, albeit more gradually. Most games now will spike my CPU temps from 30 to 65 in a matter of a minute or so which then makes my PC extremely likely to freeze or crash. I have taken every fan out and cleaned them, cleaned my radiator and blew out my case of any excess dust and wiped everything down. I have reset my Windows 10 1903; reinstalling the operating system and all of my software and drivers. But it's still happening. I have tested my CPU with various stress tests and benchmarks (3dMark and prime95 to name a few), and none of them have raised my temps like this - they only managed to raise my CPU temps by maybe 10 degrees to 40C.

My hardware:
CPU: AMD FX-9590 (not overclocked; at 4.71ghz)
Cooler: NZXT X60 Kraken with 2 fans push and 1 larger fan pull

I have started to wonder if it's a Windows setting or something since real stress tests don't seem to do it and that it only appears to happen in games.

 

[genz]

Contact surface.

Check your motherboard hasn't warped out with a tight fan mounting bracket as that can affect contact, and...
Check your case is not in contact with your motherboard as this can tap away power from the CPU, forcing your VRMs to pull more and heating up your motherboard and VRMs in the process. This can be excacerbated by gaming as when your CPU heats up it softens the board, usually making it flex more.

I had a CM case that literally hard reset the system whenever the ambient went over 40C as the mobo would wilt and as it did the back of it would touch the case and hard reset the system. I've sold systems where the same thing happened and it blew the RAM cards out. Doesn't seem far fetched to suggest it the GPU warming up could 'assist' this problem and thus when both the CPU and GPU is hot your old board bends enough to make contact with the case.

You also have basically the hottest and most inefficient performance parts on the market right now so if it would happen to any system that seems a likely one. The system with the broken Ram was Threadripper based so it takes more than average thermals most of the time.