[SOLVED] MSI RX 570 (4GB) keeps crashing

Feb 6, 2022
5
0
10
Hello. I have an MSI RX 570 for 3 years (no warranty now). Suddenly, from September it started crashing the whole PC. Every PC I tried with this card on, kept crashing. It seems to work great, until you touch it or the fans increase speed (when playing a game) and "move" the card. Then PC crashes and the monitors show some weird colors. Sometimes I get 15 crashes a day. It even crashes when the PC is idle at desktop.

I believe there's a problem with its PCI connection and it's really (really!) sensitive. Is there any way to fix this? I would buy myself a new card, if it wasn't that "scalping" mess...

Of course I tried everything, reinstalling the drivers, leaving only 1 monitor on each time, changing HDMI cables, trying an older card on the same mobo (worked great), moving from Windows 10 to 11 and the problem is still there. I'm 99% sure it's the PCI problem. I have the card screwed tightly, but the screw doesn't seem to make any difference.

Thanks.

P.S. English is not my native language, sorry for any mistakes. :giggle:
 
Last edited:
Solution
I'd try and inspect the cooling assembly on the GPU. There was a known issue with Strix cards of that era, even with GTX 1000 cards whereby if you touched the card to rectify it's sag, the temps would creep up. Logically, the person who showed that issue to me got a scolding from me, since you're not supposed to move the component while the system's running but it did bring a matter to light about the bow in a PCB and if the cooler's cooling plate was improperly seated.

As for the system you're working with, what is the make and model of the PSU you're working with and it's age? For the other system's you've tried on, do they have the same PSU as your system? You will also need to understand that the wrong drivers can also cause you've...

Lutfij

Titan
Moderator
I'd try and inspect the cooling assembly on the GPU. There was a known issue with Strix cards of that era, even with GTX 1000 cards whereby if you touched the card to rectify it's sag, the temps would creep up. Logically, the person who showed that issue to me got a scolding from me, since you're not supposed to move the component while the system's running but it did bring a matter to light about the bow in a PCB and if the cooler's cooling plate was improperly seated.

As for the system you're working with, what is the make and model of the PSU you're working with and it's age? For the other system's you've tried on, do they have the same PSU as your system? You will also need to understand that the wrong drivers can also cause you've described. As for your card, if it has any overclocks on it, back them off and see if increasing the power limit slightly helps with the issue you speak of. It could be that the GPU's saying goodbye for good.
 
Solution
Feb 6, 2022
5
0
10
Try an easy fix, back the memory frequency off 50 mhz using the and software and try that out. This is a shot in the dark, but it could work, these 570 are prone to memory issues, so this is worth a try.

I already tried that. I also did that again, after your message. I reduced the frequency (two times) from the AMD software and after two successful stress tests, the PC crashed (both times). :p
 
Last edited:
Feb 6, 2022
5
0
10
I'd try and inspect the cooling assembly on the GPU. There was a known issue with Strix cards of that era, even with GTX 1000 cards whereby if you touched the card to rectify it's sag, the temps would creep up. Logically, the person who showed that issue to me got a scolding from me, since you're not supposed to move the component while the system's running but it did bring a matter to light about the bow in a PCB and if the cooler's cooling plate was improperly seated.

As for the system you're working with, what is the make and model of the PSU you're working with and it's age? For the other system's you've tried on, do they have the same PSU as your system? You will also need to understand that the wrong drivers can also cause you've described. As for your card, if it has any overclocks on it, back them off and see if increasing the power limit slightly helps with the issue you speak of. It could be that the GPU's saying goodbye for good.

I removed the cooling system sometime in November. I even put some more thermal paste, even it didn't need it. Didn't work. My PSU is a Be Quiet System Power B8 550W. It's not trash. I think I have it for 4 years. I don't remember the PSU of the other PC, it wasn't mine, but it had a GTX 1080 running smoothly, until I installed my card.

P.S. PC crashes even if I move the tower gently for just 1 inch.
 
Last edited:
Feb 6, 2022
5
0
10
Sounds like you were right loose pace slot. Any chance the tab that holds the card is damaged? You could try a brace on the card. I personally would lay the tower on its side.its simple and could possibly work.
I have also tried putting the tower sidewards. It just decreases the average 12 crashes per day to 10. I'll try to make the card more stable somehow. Maybe it will decrease the crashes more.

I hope it lasts for some more months, until more entry level GPU cards appear on market (I'm mostly into Intel).