Graphics Card Hardware Failure

My desktop PC has an XFX Radeon 6870 that I think is failing. Last night my computer crashed, and upon restarting, I had two green vertical bars on my screen, even from time of boot. It then appeared to crash (screen black, can’t toggle keyboard lights) before getting to the lock screen.

I unplugged it and left it for the night, and tried booting it again this morning. Same problem.
Tried booting in Safe Mode and it worked, although with horizontal green lines patterned all over the screen. This is at the default 800x600 resolution.


At 1280x960, the green lines aligned into vertical bars.

I then opened my computer and booted it once more to watch the fans turn on (they did), and it booted into Windows (not Safe Mode) without problems. About 15 minutes later, it crashed again.

Has anyone seen problems like this? Is there anything I can try to do to save it, or do I just need to get a new graphics card?

More history:
I have a 650W Seasonic PSU and things have been mostly fine for the past ~3 years (so yes, it’s out of warranty). The only other crashing issue I’ve had has been once or twice a month since I got an SSD last spring. Not sure if it’s the SSD or the mobo (Asus P8P67 Pro Rev 3.1), as the BIOS loses track of the SSD until it is power cycled.

I have had similar problems to this. In my case it was a failing integrated
nvidia chip on a laptop motherboard. This ended up being a known issue with
my particular laptop. My solution was to bake the motherboard to reflow the
solder. This worked well but only held for about 6 months before starting
to go again. I may try to rebake when I have time to take off all the
plastic components.

I did it in the oven rather crudely. Let me know if you want the details
and I will dig up the settings I used. It was a quick bake. The danger of
course is baking too long but in my case it worked well. Your mileage may
vary. :slight_smile:

Personally I would just get a new card. A 5 year old video card that was awesome 5 years ago can be had in a much cheaper form nowadays.
Probably find a card for 80 bucks that performs just as good.

Or find a R9 290x for cheap that a bitcoin miner wants to get rid of.

1 Like

Thanks Janet, I might try that at some point, but I’m not all that excited to try if it’s only likely to maybe be a temporary solution.

I have a friend who can sell me a GTX 680 SC for $200 USD, which appears comparable to the 290x according to this chart. Not sure if it’s overkill, though; haven’t been keeping up with PC hardware lately.

Look at the capacitors, see if any of them have blown. If so just solder a new one on and you’ll be good to go. (I’ve actually had this happen before, and it had the a similar effect).

The other thing you could look at is if the heatsink is still sitting flat on the card, causing it to over heat. But I doubt that is the issue, but worth a shot if I’m wrong about the caps being blown.

1 Like

Yeah, I took a look at all the caps that were visible, and didn’t see anything out of the ordinary. I haven’t taken the heatsink off, so it’s possible that there’s damage hidden somewhere. The heatsink is mounted well and I don’t think it’s an overheating issue because it showed the same problems immediately on cold boot the following morning.

It does sound like it could be time for a new graphics card…if you have
the budget.

Caps can be misleading :

You can get a radeon 290x from ebay for ~$250 american, shipped.
And you can get a 680gtx from ebay for under a hundred american.

“the BIOS loses track of the SSD until it is power cycled.”

This is exactly the same problem that another friend of mine is having
with one of his new SSDs. The other one, of the same make and model,
doesn’t exhibit this problem. It’s very curious.

1 Like

Yup, use to have this happen all the time. It’s a intel chipset issue I believe.

1 Like

Thanks for letting me know! Do you happen to know if your friend is using the two SSDs simultaneously in the same computer? I ask because I wonder if it’s specific to the SSD, or specific to the SATA port or SATA chipset that each SSD is plugged into. When I upgraded from a standard HD to SSD primary + HD secondary, the SSD went into a new SATA port. My mobo happens to have two SATA chipsets (an Intel one and a Marvell one), and I’ve only used the Intel one.

eerrrchh… (see’s @toma’s simultaneous posting)

Thanks @toma! Yeah that’s what I was wondering, and was the main reason I haven’t returned my SSD because I’m not certain that it’s faulty. Interesting, I just found this page that I’d never come across before:

On January 31, 2011 Intel® announced the detection of a design error in the new Intel® 6 Series support chipset, also known as Cougar Point. ASUS has updated all motherboards with revised Intel® 6 series B3 chipsets, with every SATA port now certified safe to use and the Cougar Point support chip issue resolved.

Ok, so perhaps the solution to my SSD problem is to use the Marvell controller instead! I wish I had a working computer to try this out with right now! :stuck_out_tongue: I’m stoked to have an SSD that doesn’t show signs of failing!

Reading further into the Intel chipset issue, they claim that it only affects the performance of the 3Gb/s ports, and I’ve only been using the 6Gb/s ports. I’ll try the Marvell controller anyway, but I’m a little less optimistic now.

1 Like

I took the heatsink and fans off of the graphics card to get a closer look. No visible damage, but here are photos for those who are interested:

First, the heatsink and fan assembly:


All the through-hole capacitors look fine.

Look at all those test points (and liberal amount of thermal paste)!

And more on the back!

And here’s a bird’s eye view of downtown Capville:

no close ups of the caps?

I didn’t think it was necessary, as they all look fine. They’re all nice cylinders with either flat or slightly concave tops. No cracks, bulges, scorch marks, or anything. Do you still want photos of them?

Also, I’m assuming you only care about the large through-hole capacitors, not the SMD ones, because including both would be ALL CAPS.

I just wanna see if I can find the problem. more pics the better.

given the symptoms it would appear that the issue isn’t between the gpu and the pcie bus.
It seems like its somewhere between the gpu and video output.

If you’d like, I could bring it in tomorrow morning/lunch and you could take a look. Message me if you’d like to do that. It’d be easier looking under the microscope anyway. I haven’t been able to see anything out of the ordinary. Yeah, my guess is it’s a problem with either the GPU or the memory, given that the corrupted parts follow a visual pattern.

Turns out that I actually have the Revision B3 chipset that is the fix to the Intel SATA port bug, so that’s not the cause of my SSD disappearing… not sure what to try next, unless I’m able to get a replacement SSD.

I ended up getting a new (to me) graphics card rather than tinkering to try to fix this one. If anyone would like to take a shot at fixing it (@toma?), I’ll be dropping it off in the 3-week bin tonight. If you’re interested in fixing it, let me know and I’ll take it out of the 3-week bin and put it in the dropbox with your name on it.

All the parts are there and assembled, but you should clean apply new thermal paste before powering it because I did take the heatsink and fans off.

1 Like

I’ll give it a try. Thanks!