Status
Not open for further replies.

Yosef Marcera

Reputable
Oct 1, 2020
46
5
4,545
I built my girlfriend a PC and got a second-hand Sapphire RX 480 4GB back in November '23. The GPU is really dusty, and the warranty sticker on the screws still intact, so it's apparent that it hasn't been cleaned for years. It worked completely fine for at least 5 months until May '24. I just then found out from my girlfriend that the temps are peaking at 87–89 °C and idles at around 60 °C. There are about 3–4 times in a month where the GPU is not found on Open Hardware Monitor, and when that happens, the AMD software also says the drivers are not compatible when she tries to open up the application. Although after reinstalling the GPU drivers, it proceeds to work fine (apart from the temps).

I then finally got around to deep cleaning the GPU, repasting, and replacing thermal pads. This gave a very significant improvement in temperatures; idles are around 35 °C and max is around 65 °C. I also put in a little undervolt setting using the AMD software, and it was smooth sailing from there. Now a month later, just as my girlfriend booted up her rig, she suddenly got artifacting for 1-2 minutes, then a black screen for 1-2 minutes before the display came back to working again, yet AMD once again didn't work, so I had her reinstall the drivers once more.

It has been 5 days since this happened, and so far none of these signs have shown up again. Should I be concerned about it dying within a month or should I be fine? I've had an R9 270x die on me, but it was pretty instant (launched gtav and it went black screen and never gave a display ever again), so it's basically my first time experiencing this when it comes to gpus.
 
While the GPU itself should still be OK, I'm a bit worried about its power stages : wobbly power delivery can become susceptible to minor changes in the computer's behaviour (both the mobo and PSU send power to the card) that it would otherwise cruise through. Things can remain that way for a long time, especially now that you've cleaned it up good, or they can break tomorrow. I would revert your undervolt first in case symptoms came back.
I'd start saving up for a new GPU anyway. The RX6600 is a good deal, with double the VRAM and almost double the performance, and Polaris driver support is pretty much over now - to me, that card would only be useful on a Linux-based media center now (driver support on Linux remains strong for Polaris).
 

Imperat0r

Upstanding
Apr 19, 2024
259
86
270
Would you be so kind and tell us the system config.
Most interesting part would be the PSU.
Let's say the PSU is enough for your card and is a good quality one.
In that case you can have 3 options:
1 - the memory chips would fail because your card was abused earlier (the previous owner used it for mining)
2 - the VRM would fail
3 - GPUitself is dying.

Keep in mind that this cards (RX 4xx and RX5xx) were the best miners and this cards were really abused.
So if your PSU is a good one then for sure you have a dying miner in your system.
 
Status
Not open for further replies.