4070 ti crashing in games (overwatch and cyberpunk 2.0)

[Adam670]

Hello everyone and thank you for reading and helping

 

This is my system right now: 

 

Ryzen 7 3700x 
32gb RAM 3200Mhz

Gigabyte RTX 4070 Ti Gaming Oc

GIGABYTE X570 AORUS Elite

750W Corsair PSU

 

My monitor is 1440p 144hz
My cpu temps are "fine" around 40 degrees while idle and around 75-80 while gaming (I use the stock wraith cooler so of course I know it is hot, but no thermal throttling so... I guess is fine?
My gpu temps are nice, 36 in idle and 55-60 while gaming 

After all the information let's go to the issue, when I changed my graphics card from my 2060S to this new 4070 ti, Overwatch 2 has given me a constant crash while playing in 1440p 144hz:

 

"Your rendering device has been lost! Application closing!"

 

*Playing at 1080p is totally fine tho, no crashes at all in overwatch
 

 

I also had crashes with Elden Ring and cyberpunk 2077

games that have very little requirements to almost none work just fine 

So here is what I have tried so far: 

-Undervolting my gpu
-Going back to older gpu drivers
-Changing the resolution (that worked for ow, for cyberpunk it didn't and for elden ring I didn't test it)
-Setting the graphics on low

-checking systempropertiesperformance has enough dedicated space (16gb to 24gb max)

-SysMain is working

I also ran furmark 3 times and it seems to be alright, nothing strange, also run Heaven Unigine using 1440p and max settings and again, no problem.

I don't know if this is normal but my fans won't work until it reaches 54 degrees (I have change that with afterburner) 

And also while watching videos or doing almost nothing I could see my gpu spiking sometimes to 99% use just for brief moments, I don't think that's normal (

Please I need your help with this and sorry for my english

 

Update:

-DDU the older drivers
-Updated the BIOS
-Updated drivers for the cpu just in case

I think it may be my psu, it's a 4 year old 750w corsair RMx, which doesn't have 2 rails of 6+2 power connectors, only 1 rail with with two 6+2, the only thing keeping me away from considering this is that it should give me problems as well during the stress test, or not???
 

 

Edited October 3, 2023 by Adam670
Update

[Dukesilver27-]

I'd first try to DDU the GPU drivers and install the latest WHQL ones.

This issue is most likely caused by GPU drivers, speaking from experience.

[D2ultima]

11 hours ago, Adam670 said:

text

Firstly, I would also try DDU as suggested above, clean and restart in SAFE MODE then install the new drivers fresh.

 

Secondly, why do you have -245MHz on your afterburner? If this is what you meant by undervolting your GPU, that is not how you undervolt (though it should fix any issues with your card running at too high a clockspeed for its voltage).

 

Have you tried maxing out your power limit in Afterburner as well?

 

For what it's worth, you undervolt like this. This works for Nvidia 1000 series up until the 4000 series as of this writing.