Showing results for 
Search instead for 
Did you mean: 

Graphics Cards

Journeyman III

RX 5700 XT, some frustrating thermal problems with just... one game(almost)

So, I've picked up a "new" game. Icarus.

Now, I know for sure this game is not optimized and it shouldn't be considered great by any real means. But shouldn't be this bad me thinks.

While playing the game on minimum settings, the computer is just shutting down. Not like it goes into sleep mode or anything. It just, shts the bed and goes from playing thge game to, computer is OFF.

While watching the AMD software/tools I can see my Junction Temperatures are reaching 115. Now I understand that the car is supposed to throttle when it reaches 110 or more, but it still surges as high as 114-115 before dropping back to about 90.

Now my first thought was bad airflow so I tool both doors off my case, and that dropped the 'low end' a couple degrees, but still surging up to over 110, reading 114-115 sometimes.


Today I decided to take the 'bad air flow' theory a bit further I've set up box fan pointed at the computer, and again it drops  the low end a few more degrees but still getting surges up to above 110.

Yesterday while trying other things I tried included setting the fan up to 100%, period. So it can't come down at all.  I tried undervolting by about 10% I even tried reducing power consumption some. I tried these individually and in various combinations.

Now I had these mysterious shutdowns before but never figured out it was the graphics card until yesterday. But they were quite infrequent until this game. 

So, this has already exceeded my understanding and ability to debug. I am now screaming into the void for help. 


2 Replies
Adept II

The system shutting down hard is usually related to the power supply trying to save itself or others, in this case most likely the GPU suddenly drawing obscene amounts of power to the point that the PSU thinks it's a short circuit. Normally that should be handled by power regulation on the card and "only" make the card conk out and TDR, so maybe it's a bit of a New World vs. nVidia situation where the card design doesn't have the necessary protections.

Airflow by itself won't help with quick jumps in junction temperature, that's happening on the chip itself before even taking thermal resistance to the cooler into account. You may be able to improve things with warranty-voiding measures like new thermal paste, but that's a gamble.


Stress test your PSU, GPU, & CPU with OCCT. No need to install and it is free.

While testing keep a close eye on PSU outputs (3.3/5.0 & 12.0) plus fans speeds and Temperatures on the GPU can CPU.

Does sound like your GPU card is overheating and shutting down to prevent damage. But it could also be bad PSU outputs.

I would set everything back to "Default" to your GPU card and BIOS. Because if it isn't in Default settings you have no way of knowing if it due to one of your manually changed settings to the GPU that is causing the problem.