I recently bought a second-hand Asus Dual RTX 3060 TI OC v2 and also finally moved from Windows 7 to Windows 10. To make sure the card's fine, I stress-tested it with Furmark and everything seemed OK, it heated up to around 80 degrees C and stayed there (though the fans sounded as if it wants to take off), didn't crash. When playing Cyberpunk 2077 at high settings and some ray tracing, DLSS on, it performs admirably, with temps around 60 C. However, GPU temperature would spike at the weirdest times - e.g. during the game intro, or while I'm reading a shard (which is effectively the game paused with just some text on screen). The fans would power up, I would alt-tab to SpeedFan and see the temperature has gone up to 70, sometimes 80 C, i.e. stress-test levels. After some time on the desktop, it would calm down, I would alt-tab back to Cyberpunk and keep playing normally.
The even weirder thing happens in the other game I'm currently playing - Stellaris. After a random amount of time (once it was just after 10 sec, still in the main menu), the game would crash my entire system - first the screen would switch off with a "no video input message", then the sound would freeze, and finally the PC wouldrestart. The event viewer says the system has recovered from a x116 bugcheck, which is a display driver timeout. Analysing the crash dump also points to a graphics problem. I suspected a GPU hardware problem, but then how come Cyberpunk 2077 runs fine?!
I switched to the previous Nividia driver version, but both problems persisted. I downloaded OCCT and ran both the core and VRAM tests - no crashes, no errors found. Yesterday I switched Stellaris to fullscreen mode, turned VSync off and set the refresh to 60 Hz (same as my monitor). It hasn't crashed yet, but instead I got a GPU temperature spike, just like in Cyberpunk - out of nowhere, just browsing some in-game menus, the fans kicked into power mode and temp went up to 70 C. In Stellaris, a game my previous GTX 960 used to handle in its sleep!
I have come to believe the Stellaris crashes and the Cybeprunk 2077 temp spikes are connected. Any idea what could be causing these? I'm wondering whether it could somehow be a hardware issue with the GPU that the stress-tests could not expose - I still have a couple of days to return it.
The even weirder thing happens in the other game I'm currently playing - Stellaris. After a random amount of time (once it was just after 10 sec, still in the main menu), the game would crash my entire system - first the screen would switch off with a "no video input message", then the sound would freeze, and finally the PC wouldrestart. The event viewer says the system has recovered from a x116 bugcheck, which is a display driver timeout. Analysing the crash dump also points to a graphics problem. I suspected a GPU hardware problem, but then how come Cyberpunk 2077 runs fine?!
I switched to the previous Nividia driver version, but both problems persisted. I downloaded OCCT and ran both the core and VRAM tests - no crashes, no errors found. Yesterday I switched Stellaris to fullscreen mode, turned VSync off and set the refresh to 60 Hz (same as my monitor). It hasn't crashed yet, but instead I got a GPU temperature spike, just like in Cyberpunk - out of nowhere, just browsing some in-game menus, the fans kicked into power mode and temp went up to 70 C. In Stellaris, a game my previous GTX 960 used to handle in its sleep!
I have come to believe the Stellaris crashes and the Cybeprunk 2077 temp spikes are connected. Any idea what could be causing these? I'm wondering whether it could somehow be a hardware issue with the GPU that the stress-tests could not expose - I still have a couple of days to return it.