I think my GPU's gone bad, but I need advice

woodbeam

Distinguished
Apr 18, 2010
43
0
18,530
I've been using a Sapphire Radeon HD 7970 3GB since late 2013. A couple of days ago I started getting crashes where my monitor would be filled with a pattern of vertical colored bars. I'd seen this kind of thing before when stress testing video cards, so I immediately thought that this was a problem with the graphics card. I opened the card, cleaned it and replaced its thermal compound, and didn't see any damage while doing so, but the problem persisted. Now things have gotten worse: whenever the card is plugged into my motherboard it causes a boot loop where the system will crash before the bios interface appears, restart and repeat until powered off manually.

I'm nearly certain that the graphics card is the source of the problem; I tested my CPU's stability before the boot loop started happening, I reset my motherboard's CMOS to no effect, I tested the same setup with a different power supply with the same result and using a different graphics card does return things to normal. The only thing that's causing me to doubt is that disconnecting the CPU power supply while keeping the 7970 plugged in doesn't cause the boot loop, although the computer obviously can't boot successfully. I bought the card used to begin with, so I guess this shouldn't be that unexpected, though it was in good condition and has functioned without problems up to this point. I just want to know if there's anything more I can do before trying to get the card repaired. Thanks in advance.
 

Symbolik

Distinguished
Dec 27, 2011
45
0
18,560
Have you updated the drivers recently? If so, does this correspond to when this started happening or was it sporadic? Most of the time I see these issues, it's a multitude of things. Try a different cable between your monitor/GPU, try completely wiping the drivers and re-installing them, maybe even try a step back on the driver selection aka. try an older driver. Also, if you have two PCI-E slots, try the secondary one, clear CMOS when you switch, it helps.
 

woodbeam

Distinguished
Apr 18, 2010
43
0
18,530
I hadn't updated the drivers recently. The last update I did was for the latest package, 14.12, in December. The card had been working fine since then. Using a different cable doesn't change anything; even having nothing plugged in doesn't change the outcome. Changing drivers hasn't helped (is it even possible for them to be a factor when I can't even get into the bios?), and I get the same result regardless of what PCIE slot I plug the card into, even after clearing CMOS. Still, thanks for your advice. If you think there's anything more I can do at this point I'd like to know.
 

Symbolik

Distinguished
Dec 27, 2011
45
0
18,560
So you can't even boot into bios? Have you tried removing the card, pulling the CMOS battery with the PC completely unplugged from the PSU? Then re-installing the card. Also, cable. Did you try another cable? another type, HDMI, DVI, VGA? Try removing the CMOS battery, 5 seconds, with the card removed and then plugging the PSU back in, putting the card back in, and attempting boot.
 

woodbeam

Distinguished
Apr 18, 2010
43
0
18,530

No, I can't even boot into bios. I tried the battery procedure, the boot loop still persists. Changing output cables has no impact on the boot loop.
 

Symbolik

Distinguished
Dec 27, 2011
45
0
18,560
Sorry, I had a leave of absence from our online world. Next thing I would try is another GPU, to absolutely sure. Even a PCI one if you can get your hands on it. Also, try your GPU in a friends computer, the more scenarios you can get under your belt, the narrower the solution becomes.
 

woodbeam

Distinguished
Apr 18, 2010
43
0
18,530
Like I said in my original question, using a different graphics card does work and my motherboard does have an onboard chip which also works. I'm certain it's the GPU now, and at this point I'd just like to know if there's any way to get it repaired despite the fact that it's out of warranty. Is this thing dead forever? I'm going to contact the manufacturer about this as well, just to see. Nothing to lose now.
 

Symbolik

Distinguished
Dec 27, 2011
45
0
18,560
Hi wood, you could take off the cooler, most of them are easy to disassemble. Replace the thermal paste, I did this on my Radeon R9 270X, I used q-tips and alcohol to gently clean off the paste, then I replaced it with small, equivalent amounts where the original cooler had applied. Temps actually went down, you may want to try this, give the board and it's cooler a good dusting, check for any abnormalities in the solder points on the top and back of the PCB.
 

Symbolik

Distinguished
Dec 27, 2011
45
0
18,560
Also if it's out of warranty, you're out of luck. And yes, you have isolated this issue to the GPU, however the reason I keep commenting on possible solutions is because I thought my R9 was dead, it turns out it wasn't and I was able to continue to use it after disassembling and cleaning it. I also used a alcohol swap on the PCI-E pins themselves, give anything that looks dirty a nice cleaning.
 

Symbolik

Distinguished
Dec 27, 2011
45
0
18,560
Of course, I went through this myself with a brand new card (r9 270x) I battled the thing through every configuration I could throw at it. At the end of the day, it turns out it was my motherboard, but In your particular situation the card may just be done. All evidence points to it, you may have tried this already but if not, removal all peripherals from motherboard, only leave 1 stick of Ram, PSU, graphics, processor, and the bare essentials. If you have an HDMI cable, try that? if you haven't already.
 

woodbeam

Distinguished
Apr 18, 2010
43
0
18,530
I followed all of your suggestions to this point to no effect, and confirmed with the manufacturer that they couldn't help me as you predicted. So, in desperation, I tried the oven method. It worked. I can hardly believe it, but it's functioning now, and seemingly better than it was just before it stopped functioning. It hasn't crashed, and I've been playing a few graphically intense games that caused crashes before. I really hope this can last. Thanks again for all your suggestions.
 

Symbolik

Distinguished
Dec 27, 2011
45
0
18,560
You're welcome, you baked your card 'eh? Well, I'm glad it worked! At that point you were probably out of warranty and what else could you do. Good job! You're going to make me thinking of baking my old Nvidia 550ti again. How long and at what temp did you bake it at?
 

woodbeam

Distinguished
Apr 18, 2010
43
0
18,530


385°F for 10 minutes. It didn't look like anything had happened when I took it out, but it worked.