Video corrupt, flickering upon reboot

[Darren]

Ran into an interesting problem.

20151205_204547 by [email protected], on Flickr

My primary monitor just randomly turned into this, with a lot of the pixels flickering. My secondary monitor still worked fine and it appeared that the OS was still fully functional, just corrupt video on the primary. Upon reboot it worked but I'm getting flickering with certain shades, usually darker colors on the primary monitor still. I did a clean driver install with DDU and it did it almost immediately after. The first crash was shortly after closing Fallout 4, and the second was just when using Chrome.

Temps are fine. I removed my GPU overclocks and it's still flickering, although have yet to reinstall drivers with OC off. I did the OC with Afterburner which I've uninstalled for now. I took apart my machine yesterday to dust it, which made me wonder if I had a loose connection. I reseated the GPU and power connectors. When I reinstalled my drivers the first time, the flickering was gone up until the crash. After that the flickering has been present but yet to crash again. I'm about to do a clean reinstall again now that I've gotten everything reseated and removed MSI Afterburner.

What also concerns me is several days ago I reset my fan curve and forgot to reenable it. The beta GPU drivers had a very temporary issue of not adjusting fan speed correctly and it hit 94oC. I shut it off pretty much immediately when I realized and adjusted my controller to prevent that again. It's been fine for several days since that though, and I feel like it would have been bad from the get go if that damaged it.

I'm really hoping my 7970 isn't dead, but it kinda might be. Thoughts?

 

[voyagerfan99]

I doubt it's dead. Does it only do this after Windows has booted? If so I would roll back to a previous stable driver release.

 

[Darren]

Yeah I'm just saying screw it to using Beta drivers for now. Too unstable. It acts completely fine in safe mode and after a fresh driver install, so I'm betting it's that.

I didn't realize that I reinstalled the beta drivers after the initial crash, and subsequently had a crash. I'm guessing they're just, very Beta'y I have been on them for several days though, *shrug shrug*.

I'll give AMD credit though. They're really trying to turn around their whole driver situation with Crimson. It just needs some work.

 

[Darren]

Reinstalled older drivers and it seems fine so far.

 

[Darren]

Turns out that either version of Crimson drivers 15.11 and 15.11.1 both have stability issues and artifacting on my 7970. I used to run an overclock at 1075/1475 without issue and now I'm even getting some artifacts and throttling at stock clocks. Using the beta version of Catalyst (15.11.1) seems to be working fine so far. I do seem to be having more trouble than my friends with newer AMD cards, though.

 

[Darren]

This has started happening on Catalyst drivers also. I'm also getting the square artifacting issue I had on my first 7970 and only temporarily have fixed that with underclocking my memory core. The issue pictured above is getting more frequent. It has now appeared on both my Acer 1440p via Displayport and HDMI and the Asus 1080p via HDMI. I've done numerous driver clean installs.

I used to push this card at a fine overclock with zero issues. Ever since I had my temperatures jump up to 95oC I've been having issues that have only gotten worse. My 7970 is officially kill. I've got a friend's 760 I'll be using until I can afford a new card.

RIP.

 

[beers]

Time for the oven?
http://lifehacker.com/5823227/save-dying-video-cards-with-a-quick-bake-in-the-oven

 

[Darren]

I'm open to really whatever I can do to fix it. I feel like the 760 isn't going to like 1440p... :/

I sent AMD a strongly worded email.

 

[Okedokey]

Increase cooling.

 

[beers]

I've got a friend's 760 I'll be using until I can afford a new c

I have a 290 you might be interested in

 

[spirit]

Go with NVIDIA next time. Had far better luck with NVIDIA drivers and before you start throwing rocks at me, yes, I owned a 5870 for 2 years. Crappy drivers eventually forced me to part with it and go back to NVIDIA.

 

[Darren]

Increase cooling.

Lol.

The only increases I could make to cooling would be a water block. I have a 120mm fan right on top of the GPU blowing fresh air on it, two 120mm fans in the front, 2 fans on top (140mm and 120mm) and another 120mm on the back. Not to mention the 212 with push pull right above it. Airflow is not a concern in the least. I was playing Just Cause 3 and hitting barely into the low 60's at 100 usage.

Go with NVIDIA next time. Had far better luck with NVIDIA drivers and before you start throwing rocks at me, yes, I owned a 5870 for 2 years. Crappy drivers eventually forced me to part with it and go back to NVIDIA.

This has certainly put me off AMD cards as that's a pretty glaring problem. That said, Nvidia has also had a problem like this twice in the past according to an article I read pertaining to this issue. I'll see how AMD handles my strongly worded email. I'd probably still buy a 390 over a 970 at the moment. We'll see where cards and prices stand when I actually have money.

I have a 290 you might be interested in

If you're serious, PM me.

For the time being, I switched the BIOS on the card back to the stock 7970 as it's overclocked out of the box. I also turned off my CPU overclock and did a BIOS reset with the only tweak being clocking my RAM back up to 1600MHz as it default to 1333MHz. So far so good.

Given my issues, is there any other component that might be causing a problem? My CX600M is just now a year old, so I doubt it's that.

 

[spirit]

This has certainly put me off AMD cards as that's a pretty glaring problem. That said, Nvidia has also had a problem like this twice in the past according to an article I read pertaining to this issue. I'll see how AMD handles my strongly worded email. I'd probably still buy a 390 over a 970 at the moment. We'll see where cards and prices stand when I actually have money.

Not wanting to break any dreams, but to be honest Darren I'd be surprised if you even get a response. They have millions of customers and let's face it, they hear the same old from people like yourself who are having problems all the time. I don't think they'll do anything. What did you say to them?

Sure, NVIDIA has had their issues, but generally I've had more luck with them and let's not forget NVIDIA update their drivers every few weeks. When I had a 5870 I was lucky to receive driver updates every quarter. If you get a bad NVIDIA driver chances are it won't be for long, with AMD, in my experience at least, it's longer before you get an update.

 

[Darren]

I don't expect anything either, but I wanted to try.

I basically said I've done extensive testing and am getting problems under identical hardware/software configurations that I had tried before the temperature episode. I need to try contacting Sapphire too.

AMD's been pumping out a ton of driver updates the past couple months and fixing problems. They're transferring over from Catalyst Drivers to Crimson Drivers, so it's a bit of a transition period. The fan issue I had was fixed within a day or two I think.

 

[Darren]

Still early to tell, but it seems happy now. I switched back to vBIOS one with the onboard switch which sets it to 950/1425 which are the default 7970 clocks. I had manually set the clocks to 1000/1425 with Afterburner while on vBIOS 2 (factory overclock). When I was on vBIOS 1 Afterburner was still running those clocks for a day or so before I noticed. Even then it still seemed fine. Uninstalled Afterburner and now at 950/1425. Yet to have any issues or artifacts. *knock on wood*

I also reseated and further dusted the card when I did the vBIOS switch. Although I had done those before too and had the video corruption both before and after. If this stays happy for the next few days, I'm going to try clocking it back up to 1000/1450 and see how it responds. The card is the "Boost" version and came new with 1000/1450 out of the box, so I'd like to run it at those clocks at the least if I can. If those are fine I may try bumping it up again as I ran it at 1075/1475 for a couple months with no trouble and no temperature concerns.

Is it possible the vBIOS was corrupted or is it probably just playing nice since it's downclocked? Also, does vBIOS mess with voltages at all or just clocks?