AWT 6. nov. kl. 18:31
GPU Problem?
My GPU crashed with a loud repeating beeping noise presumably coming from MSI Afterburner (thru speakers) yesterday, with the monitor losing signal (but system seemed to continue working according to LED lights etc.). I quickly shut it down via held power button.
Today, my computer only starts occasionally (and sometimes BIOS reports system instability, although when I then restart with CTRL+ALT+DEL, the system will boot normally.
However, whenever I start a game, it crashes within seconds (again monitor signal lost, system seems to continue operating otherwise but without video output).
The GPU, a Zotac RTX 3080, only has partial RGB lights or sometimes no RGB lights.
I removed all the RGB software, tried a "cold reboot", installed latest graphics drivers, but to no avail. Sometimes the system even drops video signal while in idle / watching videos on YouTube (sound continues despite video signal drop).
I just noticed the PCIe PEX Errors Recovery counter in HWmonitor is rising continuously - it currently stands at 953 Errors after about 15-20 minutes since last boot. No other PCIE errors are reported though.
I made sure the card is inserted properly, and the power connectors are fully inserted.
You can find my full specs in my profile.
Should I assume the GPU is dead, or could this be the mainboard (which is clearly aging), or the cheap PSU I bought a couple of months ago? (I faced a similar situation back then - my old PSU died, and I didn't want to buy an expensive PSU only to realize it's in fact the mainboard or GPU that's dead, so I got the cheapest one to see if the system still works. I didn't intend to use it for much longer, but it would seem odd for it to go bad THIS quickly even if it's rubbish - time to buy a better one anyway I guess).
I had issues where the card wouldn't be recognized/start until I fiddled around with it a bit (pushing it inside the slot harder, needing to disconnect/reconnect power repeatedly, etc.) when reinserting the card ever since I attempted to run it via a riser cable, so I guess some contacts could be bad. Any ideas? Should I take it apart? It was always running at 83-87° under load...
Thanks in advance for any advice
Sidst redigeret af AWT; 6. nov. kl. 18:33
< >
Viser 1-3 af 3 kommentarer
smallcat 6. nov. kl. 19:05 
Dont you have a spare card to test on ? Any errors logged in Event viewer .it s clear the PSU is not great at all but cant blame it for now
Sidst redigeret af smallcat; 6. nov. kl. 19:07
Guydodge 6. nov. kl. 19:21 
sounds to me like the riser cable may have screwed something up try the other Pci slot.
and check the contacts on the gpu that go into the pci slot for damage.
.
Sidst redigeret af Guydodge; 6. nov. kl. 19:24
AWT 6. nov. kl. 19:30 
Won't the other PCI slot slow down the card massively? It's already a PCIE 4.0 card running on PCIE 3.0, and switching to the 2nd slot would reduce it to PCIE 3.0 x8 instead of x16.

I will be meeting a friend on Monday to borrow a spare GPU for further testing (leaving for a 3 day trip tomorrow).

UPDATE: I tried connecting the GPU to the 2nd PCIE slot (ChatGPT says loss should be negligible), but it doesn't get detected at all if connected to 2nd slot. I tried reseting BIOS (as GPT suggested I should do this to retrain lanes), but it didn't work.
It now worked for a short time (I played 2 rounds of World of Warships, then it crashed in the menu). Again just signal drop.


Thanks
Sidst redigeret af AWT; 6. nov. kl. 22:17
< >
Viser 1-3 af 3 kommentarer
Per side: 1530 50