Jul 20, 2018
Hey guys, any time I find a bizarre solution, I'm going to start posting here to help others out.

Alright so today I got in two used XFX RX 480 8GB GTR's, which are supposed to be fairly high quality as far as the 400 series goes.

So I go to put these cards into gaming rigs and both of them overheat instantly when gaming.
At idle they were sitting at 45-60c (fans off), but the second that they entered a 3D environment, even relatively basic environments like Tomb Raider (2013) they would both instantly skyrocket to 85-93c with no temps in between.
In addition, Furmark was basically a killshot, as both would instantly overheat to the point of turning off the system once it was turned on - so quickly in fact that you wouldnt even see the fur load in.

Okay so naturally, it's an overheating issue, so the first suspect is thermal paste. I tear the cards down and yeah, the factory paste looks dry and generally pretty low quality. I clean it up and replace it with Noctua NT-H1 and throw them back together and test.
Basically this did very little. temps went down to 83-90c instantly rather than 85-93c, but the max operating temp of this card is 83c, so naturally this still wasnt acceptable.

So from here I browsed forum after forum and saw a handful of people with the same issue but no solution. I tore down the cards and repasted probably four times, wasting a ton of Noctua paste...

BUT I FOUND THE ISSUE.... and it appears to be an issue straight from the factory.

So next to the CPU die and on the VRAM area there was a rather considerable amount of an oily substance on both cards - I've seen it before, its an oil released by some thermal pads used on VRMs. I was kind of out of options, so I removed the VRAM heatsinks and found that about half of the VRAM chips had a "pool" of oil on top of them, basically sitting between the chip and the thermal pad.
I used a qtip and 99% rubbing alcohol and removed this oil. I also used dry q-tips to try and dry as much of the oil as I could off the cards.

After doing this, I did general cleaning (getting dust off the thermal pads, which were still in good condition) and I reassembled the cards with more Noctua paste.

Following this, temps dropped drastically. The cards were idling at 35-43c and under load were maxing out at 65-70c, with a gradual incline over time as I would expect them to.

if repasting doesnt solve overheating issues on a graphics card, the VRAM may actually be what is overheating - maybe the oil I found was interfering with proper cooling, but this seems relevant and I havent heard a single person mention this.

Anywho, if you have any input on this, let me know, as I've never seen or heard of it before. Hopefully this saves someone from thinking their card is just bad.