post #1 of 1
Thread Starter 
I got this used HD7970 a few days ago....
I finally thought I had the perfect HD7970. It's got the V1.00x PCB, voltage unlocked, 10 phase VRM, Hynix memory, good ASIC, etc.

Initially I did not have any problems (although on the very first boot I believe it got stuck once loading the AMD driver, just after the WIndows boot splash screen.)
It worked fine no artifacts. So I proceed to overclocking of course. I find out it is voltage unlocked and become excited.
First I do a few tests without any voltage increase just to see how it works. I get to about 1150 MHz in Unigine Heaven on the Core, best I have ever seen on a Tahiti card that I've tried.
However, I notice the VRM is toasty. Hits 100C with ease, and increasing slowly beyond that if time allows it to. Ambients were high though.
But more strangely, when the GPU is running IDLE, the VRM still runs at about 70C. Not good.

So from here on out I start trying loads of things with the Cooler, I also have two different Arctic Accelero Xtreme coolers, and I tried all sorts of stuff.
In this process, I also ran the VRM without heatsinks once. Surprisingly, the temps were not that much higher than with the large ASUS heatsink. Particularly during idle it did not make much difference.
So first question, was the VRM faulty? I mean it is a 10 phase VRM and it gets to 70C just running idle in Windows? What the?

Somewhere in my testing, I tried OCCT once to test the stability further (with VRM heatsink on of course). I think the clock was at 1100 MHz on the core. It worked well for the first minute but the VRM temp was climbing very fast.
As it crossed the 100-105C mark, the Vcore started dropping slowly. Understandably, otherwise the thing could blow up. Of course, it would not be long until the card would hang. But it did not do that, it crashed in a violent way.
The entire PC hard locked with black screen. Not good. But after draining the power it came back. I thought it would be OK. From here on I tried some more testing but no OCCT or Furmark anymore.
I ran Crysis 3 Welcome to the jungle without issues.

My idea was to use the Arctic Cooling Xtreme on the core, to reduce the core temperatures compared to the DCU2 cooler, which is kind of weak. Because lower core temperature means less stress on the VRM.
Secondly, I wanted to use the backplate as extra VRM cooling, by placing a thermal pad between the back of the VRM and the backplate. I did not have a thick enough thermal pad however.
This is where everything went completely wrong. After this backplate experiment, the entire screen was tinted and there were artifacts all over the screen.
Interestingly, the UEFI boot screen and also Safe mode were unaffected. At first I thought it might be the monitor cable or something, but the fact is worked in safe mode and not after loading the driver meant that it had to be the hardware.
After this I tried re-cleaning and repasting the card but nothing helped. In fact, the poor thing has gotten worse and worse.

If I try the card in my Gigabyte motherboard, the UEFI splash screen is tinted yellow. You can barely distinguish the ultra durable logo from the background. When it loads into windows, the entire screen is black with some lines running through it.

RIP I guess.

So.... should I try my heatgun on it? I'm waiting for my IR temp gun to arrive to I can reflow it with some degree of precision.