Question Unidentified problem with Gigabyte RX 580 8GB ?

Kian960

Honorable
Mar 3, 2017
6
0
10,510
0
This might be long so please bear with me.

So, a week ago. my RX 580 got splashed by drinking water, it was an accident as my brother hit the cup by accident. The gpu was in vertical position so the entire thing got showered with water, the system is safe. However, not the gpu. Garbage display shows up, I immediately plugged the power off my computer.
What I did was, opened it up, sprayed a lot of isopropyl alcohol on it, the entire gpu. And waited for a day. Plugged it back in on my pc and it runs, but there was no display. I can hear the windows logon sound, but no display.

So I thought it was already dead. I waited 2 days later, I tried opening it again because I couldn't buy a gpu and wanted to see one more time if it's the same case of just displaying nothing but black screen.

Miraculously, I see windows logo, my entire system, now I didn't run anything immediately as I was cautious that the gpu might die again.
So first thing I did was, opened GPU-Z, everything seemed to be normal until the memory section.
Instead of displaying GDDR5 (Hynix) memory, it instead displayed: GDRR5 (Autodetect). Which was weird because I can clearly remember it was hynix, so I opened up the gpu again, and tried blowing the sides of the gpu memory chips, and even used a thin card and slid it across the sides of each gpu memory chips(I have no idea why I did it I just think it could do something, I am no repair guy, just a regular dude who can't buy a gpu at the moment). And tried it again, and tada, GPU-Z now detects the memory as hynix.

So I tried playing a game, world of tanks. Not too long in the game, around 2 minutes or less the pc crashed. I was scared because the gpu maybe dead for real, but it still restarted back to windows(Thank goodness). I tried it again multiple times and still the same(I even got deprived of credits and EXP on wot because I kept on leaving the game). So, this time, I tried undervolting the gpu voltage, it didn't help much.

However, this time I set the power limit up to 50%, it lasted a bit longer, and I mean by longer it went from 2 minutes before crashing all the way to 1 hour and 12 minutes before crashing.

it made me wonder that it could be something related to gpu VRM or mosfets?? Again, I don't have much knowledge when it comes to this but, it usually lasts long with the power limit being to 50% but it still crashes.
And, did I forget to say that the gpu temp is always at 90c?
Yep, the gpu was actually staying at that temp whatever the power limit is, so long as it hits 90-100% usage it goes to 90c.

Originally, my gpu always maxed out at 73c at full load. My thermal paste didn't change either. It was a Thermal Grizzly Kryonaut.
So please, to everyone that can help understand what's going on, please tell me because I am out of options here, I don't know if it's still repairable or still can be saved.
If there's any questions regarding the gpu, I'll do my best to answer them.
Right now I am using the pc with the same gpu as I make this thread.

I'll post my entire system spec for this as well:

AMD Ryzen 5 3600
Corsair vengeance RGB 3200mhz 16gb ram
samsung evo 250GB SSD
HGST Ultrastar 2TB HDD
Gigabyte B450M DS3H
Cooler master MWE 600 bronze
Gigabyte Gaming RX 580 8GB
 

Eximo

Titan
Ambassador
1-day probably wasn't long enough to wait, and the alcohol was a good idea. Your information about the higher temperatures is a bit disturbing, might indicate some internal sensors have stopped working.

It is possible there is still liquid underneath the memory chips and GPU package. Toss it in an oven at 225F (just over 100C) for a while with the heatsink off of it. (This is not a reflow procedure, you want the card to be dried out completely, which running the card may not accomplish)

Re-apply thermal compound and thermal pads and test again. Clean the PCIe connector contacts. It might not come back, but you don't have too much to lose. GPUs prices are finally back down, so an RX6600 isn't going to cost you a kidney.

Check the motherboard for any dust or debris that the water may have gotten into the nooks and crannies.
 
Reactions: Kian960

Kian960

Honorable
Mar 3, 2017
6
0
10,510
0
1-day probably wasn't long enough to wait, and the alcohol was a good idea. Your information about the higher temperatures is a bit disturbing, might indicate some internal sensors have stopped working.

It is possible there is still liquid underneath the memory chips and GPU package. Toss it in an oven at 225F (just over 100C) for a while with the heatsink off of it. (This is not a reflow procedure, you want the card to be dried out completely, which running the card may not accomplish)

Re-apply thermal compound and thermal pads and test again. Clean the PCIe connector contacts. It might not come back, but you don't have too much to lose. GPUs prices are finally back down, so an RX6600 isn't going to cost you a kidney.

Check the motherboard for any dust or debris that the water may have gotten into the nooks and crannies.
I don't have an oven.. but there is a shop I can go to, apparently it's a repairs shop with a infrared meant for reballing gpus, you think this will help drying out the gpu?
As for the temps.. yeah I know, it's weird. the idle is literally 51c now, when it was around 45c before.
 

ASK THE COMMUNITY