Black screen when launching games, but not when stress testing...

[Psych0]

Hello everyone, I am building a new system and I am still waiting on my GPU (RTX 3080) to arrive. I have found similar problems, but not as specific as mine... so sorry if this is a duplicate thread but I haven't found one that's the same. The problem I have with this system is the following... Whenever I launch a game, I get a black screen (no signal from gpu), sometimes during the loading screen or after 15-20 sec in the game. I can still hear the sounds in the game, and after 10 seconds the black screen has occurred I can hear the notification sound of the windows like a new hardware being plugged in / detected. At this point I have to restart my pc to regain the display. After the restart I can see the notification that the display drivers have crashed. I have tried stress testing the system with Aida64 Extreme and it goes fine without freezing / crashing / black screens. I have also tried Memtest86 and it shows no errors / freezing even after 2-3 hours of running. I can leave YouTube on for hours and it doesn't crash. I am waiting on my RTX 3080 to arrive, and until then I am stuck with this old GPU. To make things even more complicated, the GPU works like a charm when I plug it in the old computer I have... no crashes, nothing.

Things I have tried so far:

1. Stress testing with Memtest86, Aida64 Extreme (All good)
2. Launching different games (all result with black screen)
3. Turning off XMP, Precision Boost Overdrive
4. Removing one RAM stick, booting with one, switching slots and what not...
5. Trying both PCI-express lanes
6. Trying 3 different power cables for the GPU
7. Reseating the GPU countless times to make sure the double lock is properly engaged
8. Removing BIOS battery and clearing
9. Clean install windows
10. Tried different SSD's and windows versions
11. Tried cleaning drivers with DDU and installing older drivers
12. Tried undervolting and downclocking the GPU

I Q-Flashed the bios when I received the motherboard in order to prepare it for the CPU (Ryzen 5900x). Everything went great, and the bios version is F31o (AGESA Version 1.1.0.0).

My system:

MB: X570 Aorus Elite (F31o bios)
CPU: Ryzen 9 5900x
Ram: 2 x 16gb 3600Mhz G.Skill TridentZ RGB 16-19-19-39
GPU: Asus STRIX GTX 970 Direct CUII 4Gb
SSD: Kingston 240Gb (just for testing purposes)
PSU: Evga Supernova GA 850W Gold Plus
Cpu Cooling: Corsair iCue H100i RGB Pro XT
Monitor: 2 x Dell U2719D
Windows: Win10 Pro x64 (latest update)

Any help would be greatly appreciated, as I want to use the new system with my GTX 970 until the new GPU arrives. Thanks!

 

[joemama]

Seems like a GPU problem, have you tried using furmark to stress the GPU alone? Also check the temperatures while stressing the GPU to see if anything's weird.

 

[Psych0]

Seems like a GPU problem, have you tried using furmark to stress the GPU alone? Also check the temperatures while stressing the GPU to see if anything's weird.

I haven't tried furmark on the new build. But when I use the gpu on the old computer it plays the same games without any trouble... GPU temps seem to hover around 60-70 C

 

[Psych0]

**UPDATE**

I tried Furmark and it crashes. Sometimes after 30 sec sometimes after 10 sec, it's random.

 

[Psych0]

***** UPDATE *****

After flashing the latest BIOS F33a with AMD AGESA ComboV2 1.2.0.0 , it seems like it's not crashing that fast now. After a bit of testing I got some results:

1. When the GPU is plugged in the main PCI-Ex 16 slot it seems that Furmark can run for 5-6 mins before crashing to a blackscreen. If I downclock the GPU (90% max power, -45 mem and core clock) it seems that it can run for 17-20 mins before crashing.
2. When the GPU is plugged in the second PCI-Ex 4 slot it seems that Furmark can run for 45-50 mins before crashing to a blackscreen.
3. I managed to actually run a game for 5-6 mins before crashing. That wasn't possible before the BIOS update as it would crash in the first 10-15 seconds.
4. If I overclock the GPU the crashing seems to be in the same 10-15 seconds as before.

Is it somehow possible that the GPU is getting a bit more power than it can handle on this motherboard? Is there a way for me to increase or decrease power limits on the PCI-ex slots? I mean if it was a hardware issue, surely the timing of the crashes would be consistent no matter the BIOS versions or downlocks of the GPU....