• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

SOLVED Did my 7970 just die on me?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

fdrph

New Member
Joined
Jun 23, 2015
So here I was all happy playing elite dangerous when suddenly the computer freezes and locks completely. My first monitor was black and my second monitor had very fine stripes of white red etc vertically. The card was managing fan speed on its own and I thought at first it just overheated and shut down to protect itself or something (even though it shouldn't have overheated since it was fine regulating its fan speed for all the other days I played..). This isn't anything a reboot won't solve, especially in AMD land, I've seen this happen.

But to my surprise when I reboot I am greeted with this:
accce35b08.jpg


Now THIS is worrying, red bars / artifacts out of the operating system smells like a dead card. And when the Win 8 boot started it just showed black. Now this would completely ruin my summer break (since my last exam was TODAY when this happened.. tough luck)

So from here I took the card out and let it cool (thinking the cause was overheating) popped it back in exact same problem, bars at bios screen and black at win 8 boot.
So I changed the main GPU to the integrated one on my 4790k and went into win 8 with the intent of finding if windows recognized it. All while plugging the 7970 to a second monitor. my device manager showed two "Microsoft basic display adapter" with one being yellow (the 7970), GPU-Z also detected it like a basic display adapter and on the second monitor there were the SAME red bars.

Then I remember the only mod I made to this card, installing an UEFI bios like over a year ago, so here I just switch to BIOS 2 (it has a switch :3 )

Well some progress was made since now it boots into windows 8 (doesn't show just black) without any integrated GPU, all by itself.
But the red bars are still here and it still is detected as a basic display adapter ( not yellow this time).

3f1c97a3f5.jpg
be4e33244c.jpg

Curiously the bars don't appear in screenshots, but trust me they are still here ( why is this? maybe bars are added at the end of the rendering cycle when it shoots the data to the monitor and hence windows isnt picking those bars up, I dunno, it still smells like hardware damage and perm dead card, but I have hope)
If I try to install the catalyst drivers, when it installs the display driver it just stops outputting ANY signal and turns black and I have to reboot. Also the fan gets quieter.
So from here I am just lost.... I don't know if I should try to flash the BIOS 1 into the default BIOS the card came with (should be on the web somewhere) or just call it a day and cry myself to sleep since I don't have money to buy a new card right now. I really need advice.
 
Last edited:
pull power cord and hold your start button to discharge all power..
see if card is still messed up.

unfortunately this doesn't sound good to me.

you may even want to pull the card for a period of time and let it sit outside the mobo
 
Yeah I did that SF101, I let it cool in the night air for awhile! :p also pulled cord.. but...

... this is REALLY weird, I found my original BIOS rom of the card and flashed BIOS 1 with it, the problem was still there, NOTHING changed. I was just looking at the same red vertical bars with no hope, my wallet on my hand while I browse sites to buy a 280X when suddenly the bars turned thinner for no reason! also they would change when I passed my mouse over, anyway, just weird artifacts. after awhile of looking at this weird change I rebooted and........... THE BIOS SCREEN HAS NO LINES... wait a lil bit... WIN 8 HAS NO LINES.... try to install the catalyst drivers.... IT DIDN'T JUST TURN TO BLACK .... and it just worked suddenly.
81695b39d4.png
Now... I have no idea how electronics work, but right now I really wish I was an AMD engineer, because this isn't normal, I'm afraid it was a warning and will happen again for good, for ever o.o so I really wish I had a fundamental understanding if how this could happen... I only have theories in my head, and many of them involve this being a "warning" it will die soon.
What kind of damage can be done to a GPU where:
-this damage is done by itself
-it introduces artifacts in the output but not in the rendering
-the damage is reversed with time
-could be caused by overheating (since I was playing elite at the time)

It really was weird, I have NEVER seen something like this... I'm sorry for wasting a thread, but it could die soon, if it does I'll just post a quick sentence saying it did.
Anyway, this should be solved now, I guess. I'm still shocked though..
 
Now... I have no idea how electronics work, but right now I really wish I was an AMD engineer, because this isn't normal, I'm afraid it was a warning and will happen again for good, for ever o.o so I really wish I had a fundamental understanding if how this could happen... I only have theories in my head, and many of them involve this being a "warning" it will die soon.
What kind of damage can be done to a GPU where:
-this damage is done by itself
-it introduces artifacts in the output but not in the rendering
-the damage is reversed with time
-could be caused by overheating (since I was playing elite at the time)

It really was weird, I have NEVER seen something like this... I'm sorry for wasting a thread, but it could die soon, if it does I'll just post a quick sentence saying it did.
Anyway, this should be solved now, I guess. I'm still shocked though..

This is a sure sign that the GPU Core is becoming desoldered/disconnected from the PCB/circuit board.

Over time the repeated sessions of heating (during gaming or other heavy use of the GPU) and then cooling (when not in heavy use, i.e. just sitting at the desktop in 2D mode) will tend to cause a GPU core to become disconnected from the PCB that it is soldered to. This is caused by the solder breaking down when "cracks" or other small gaps form in the solder joints, causing loose or bad electrical connections. The result of this is visual artifacting (such as the red bars you saw on your monitor), or an outright failure of the unit leading to no visual output to the monitor at all (when there are many broken solder joints). Also, over time from exposure to these heating and cooling sessions the PCB can "warp" out of it's original shape (usually just slightly), which can also cause the solder connections to be broken.

This happens to both AMD and Nvidia graphics cards, so it's not limited to one manufacturer or another.

I've seen these symptoms before in some of my own graphics cards.
 
Last edited:
Yeah I was also thinking about heat causing warping in components and causing artifacts but your description is very nice and I appreciate the explanation.
I've heard of this before, this kind of damage to solder and the technique of "reballing".
But there's no way to mitigate or avoid further damage except not using it. Therefore I don't think I can do anything except wait and observe.

Furthermore, there's a PC repair shop close to me that has equipment for repairing GPUs, but they work on laptops mostly, I don't know if they have any experience with dedicated GPUs or if this can be fixed by anyone really.

In any case, thanks for everything :p
 
If your card is working now i recommend creating a true fan profile in Afterburner so it never gets too hot again.

I always setup my fan profile so if the card hits like 75+*C the fans ramp up and cool it down & it seems to have served me well for years doing so.
 
That's exactly what I did, I never let it above 70C now even if it's super loud.
 
Here I am repeating third hand information again....

I recently read a post around here about "baking" a mobo to get the solder back where it is supposed to be. I have no idea if this is a viable option in your case or what the risks may be, but it was discussed as a sort of" last resort before the trashbin " option.
I guess all you do is treat your card like cookies and throw it in the oven. Now I read that some solder melts @ 150 ( Fahrenheit I assume... Don't be shy with the google) but yours may be different (although I can't imagine it being much more, or less due to the typical load temps. Leave it in there until the solder melts then pull it out and let it cool. Then maybe do the fan profile thingies.
 
Yes, reflow it UNLESS you still have warranty.
Then RMA it.


If not, we have a thread about reflowing your GPU here. ;)
 
Warranty ended, but it is working for now, if it gets these artifacts for good, I will try that. But don't forget the store nearby I mentioned that specializes in resoldering boards and GPUs, maybe they can do it more professionally, I don't know if they work with dedicated GPUs.... we'll see how long it lasts.
 
Back