Question Computer black screens after installing new (2nd hand) gpu.

[stefan999999]

I upgraded my pc a few days ago with a new ssd, ram and second hand graphics card and started with a completely clean install of windows. I ran 3d mark and furmark to check if the system, mainly the graphics card, was running as expected and it was. It got a reasonable score in 3d mark steel nomad (39xx with 41xx being average) and furmark completed without any issue. Then suddenly when gaming for a bit the screen went black, but the graphics card logo was still lit up (it has an rgb logo as those gaming gpu's have) and the fans were still spinning. After waiting for a bit and still getting no display, I held down the power button and booted it back up and it seemed to work fine again, until it happened again. However this time it wouldn't turn back on so I tried to reseat the graphics card and the power cables going into the graphics card and when I pressed the power button there was a loud bang and a bright flash. I disconnected everything and took out the power supply and it was rattling. I got the item out by shaking it a bit and it contained a part number which I took an image of. Then I put in a new psu after looking at the tier list and everything still worked luckily (first tested things with the old graphics card). However, after running the computer for 2.5 hours it happened again: screen went black but light was still on. I shut down the pc again but was too afraid to start the pc again in fear of another explosion, so I put the old graphics card back in and am now writing this post.

Furthermore, I do believe the old PSU was already faulty since I always had frequent lag spikes in a lot of games which I had for both GPU's with the old PSU, however when I changed the PSU both GPU's had no lag spikes anymore.

Also, what I noticed is that the old graphics card (5700xt) has little leds in front of the pcie ports that light up when the pc is plugged in. I saw these lights were also present on the new graphics card (6900xt), however they don't light up, not when the pc is plugged in, nor when the pc is on.

I am not sure what to do next so I'm consulting here.

My specs are:

• AMD Ryzen 3600
• G.Skill Aegis F4-3200C16D-32GIS
• Samsung 990 Pro 4TB (without heatsink)
• New graphics card: ASUS TUF Gaming Radeon RX 6900 XT
• Old graphics card: ASUS ROG Strix Radeon RX 5700 XT
• New PSU: Corsair RM1000x
• Old PSU: Asus ROG Strix 750W Gold (blew up)
• Mobo: ASRock B450M Pro4

My bios is from 9-12-2019, P3.90

I am running windows 11 64-bit and running the latest drivers from the AMD adrenalin edition
For what it's worth, the last crash where I was too afraid to go anything further happened when trying to boot COD Cold War
The pc doesn't have a minidump, but I've provided the application and system event viewers.
What I also found strange is that it says that the system (unexpectedly) shut down at 19:06:15 on 18-8-2024, however while I wasn't behind my pc I don't remember it shutting down between 17:00 and 19:44 when that last crash happened.

Event Viewer Data

 

[Aeacus]

and second hand graphics card

Used hardware is sold "as is" and there is 0 expectation for it to work like brand new hardware. Nor is there any warranty with used hardware.

So, when everything works fine with old GPU but not with the used 6900XT, then the issue is with the 2nd hand GPU you bought.

but was too afraid to start the pc again in fear of another explosion

Corsair RMx is good quality PSU and it won't blow up. At best, PSU itself may die with small "crack", but chances that PSU fries everything else - is low. Moreover RMx has working protections in it and those protections would shut the PSU down, before letting the PSU to go "boom", like it was with your Asus unit.

Tough, Asus unit is great quality and it killing any other hardware is minimal. That you also saw, all other hardware you have, survived. As of why it went "boom", difficult to tell. But there are lemons, even among the best. So, this is what i think happened. Better to RMA the Asus unit, so that Asus can inspect it and see what went wrong.

 

[Lutfij]

Welcome to the forums, newcomer!

started with a completely clean install of windows.
Did you install the OS in offline mode, later installing all relevant drivers in an elevated command?

What BIOS version are you working with for your motherboard?
https://www.asrock.com/mb/AMD/B450M Pro4/index.asp#BIOS
You have BIOS updates pending. Speaking of BIOS, can you remain in BIOS indefinitely?

I don't see any mention of DDU. Use DDU in Safe Mode, to remove all GPU drivers(Intel, Nvidia and AMD) then manually install the latest GPU driver sourced from AMD's support site but in an elevated command.

 

[stefan999999]

Used hardware is sold "as is" and there is 0 expectation for it to work like brand new hardware. Nor is there any warranty with used hardware.

So, when everything works fine with old GPU but not with the used 6900XT, then the issue is with the 2nd hand GPU you bought.


Corsair RMx is good quality PSU and it won't blow up. At best, PSU itself may die with small "crack", but chances that PSU fries everything else - is low. Moreover RMx has working protections in it and those protections would shut the PSU down, before letting the PSU to go "boom", like it was with your Asus unit.

Tough, Asus unit is great quality and it killing any other hardware is minimal. That you also saw, all other hardware you have, survived. As of why it went "boom", difficult to tell. But there are lemons, even among the best. So, this is what i think happened. Better to RMA the Asus unit, so that Asus can inspect it and see what went wrong.

Yeah I understand the risk associated with buying second hand, however the new graphics card did run fine on the pc from the guy I bought it from, and also runs fine for a few hours on my own pc, so I feel it's a bit too rushed to assume the GPU is broken since there might be something I'm missing. For example I've read on other forums about turning on and off HPET, and someone wrote about turning off MCE, while someone else said that even if the GPU is crashing at some games, it could sometimes be resolved by underclocking the GPU which I all plan to do when I have the time again.

However, I'm still asking here since there might just be something I'm missing.

 

[stefan999999]

Welcome to the forums, newcomer!

started with a completely clean install of windows.
Did you install the OS in offline mode, later installing all relevant drivers in an elevated command?

What BIOS version are you working with for your motherboard?
https://www.asrock.com/mb/AMD/B450M Pro4/index.asp#BIOS
You have BIOS updates pending. Speaking of BIOS, can you remain in BIOS indefinitely?

I don't see any mention of DDU. Use DDU in Safe Mode, to remove all GPU drivers(Intel, Nvidia and AMD) then manually install the latest GPU driver sourced from AMD's support site but in an elevated command.

Thanks for your reply.

I installed the OS with an internet connection, but now that I think back I plugged out the ethernet cable when it prompted me for an Microsoft account so I could create a local account and then plugged the cable back in. I'm not sure if this caused the install to continue in offline mode and if the drivers are even installed after that point.

The only drivers I installed where the GPU drivers by just installing AMD adrenalin edition and installing the drivers for the GPU from there.

Do you think updating the BIOS will have any effect? From what I know it is mostly for adding support for CPU's and adding security updates. I do see an entry with "Improve GPU compatibility for GeForce RTX 40 series." however this is not an RTX 40 series, it isn't even an Nvidia GPU. Besides I have not yet tested if I can remain in the BIOS indefinitely, I'll see if I can try that.

Also, why use DDU on a clean install of windows? It shouldn't even contain any drivers then yet right?

Also with elevated command you just mean executing it as administrator right?

Don't know if this is relevant but I installed windows 11 on the machine.

 

[Aeacus]

For example I've read on other forums about turning on and off HPET

Disabling the High Precision Event Timer (HPET) is a troubleshooting step that has been suggested in the past to address certain gaming issues, particularly related to micro-stuttering or frame rate inconsistencies. However, what you are experiencing, is not micro-stutters of FPS issues, instead complete loss of video signal. Disabling HPET won't give you any benefit at all.

someone wrote about turning off MCE

Multi Core Enhancement (MCE) is Intel CPU only. You have Ryzen, so, you do not have MCE option in BIOS. For Ryzen, there is PBO and Auto OC, but how do CPU core frequencies define what GPU is supposed to do? 🤔

it could sometimes be resolved by underclocking the GPU

It could or it could not. Won't know before you try it.

Yeah I understand the risk associated with buying second hand, however the new graphics card did run fine on the pc from the guy I bought it from, and also runs fine for a few hours on my own pc, so I feel it's a bit too rushed to assume the GPU is broken

You do not know what kind of abuse 6900XT has seen before. And just because it worked a bit, doesn't mean GPU is sound. Since if GPU would work normally, you won't be getting the black screen issue.

You've already established that system works fine with 5700XT, right? Whereby when you change the GPU to 2nd hand 6900XT, issues start happening, right? Since only thing that you replaced was GPU, the issue is with new GPU. Else-ways, you'd be getting the same issue with 5700XT as well, IF the issue would be somewhere else in the system (e.g with CPU or MoBo).