R9 290X's broken?

[d3rrial]

Hey there, was wondering if any of you know anything that could help me:

Short version: My 2 R9 290X cards show artifacts (horizontal lines, looks a bit like tearing), then go into black screen mode around 10-20 seconds after the horizontal bars appear, system freezes up when I move the mouse after that.
Same is true for both cards, tested individually. Did some minor overclocking, stayed well within temperature limits set by AMD.

Long version:

I was doing some scrypt mining yesterday on my GPUs, everything was fine, however, as is usual when I am changing settings on the GPU, when i close BFGMiner, my PC BSODs.

So I was mining at a chilled 1ghz core clock, 1.5ghz memory clock, powertune +20% temps stayed between 70°C and 96°C (which I set as overheat temp, so if it hit 96°C it would ramp up the fans to 100%) temp cutoff was set to 100°C at which point bfgminer would have killed the thread on that GPU (which didn't happen)

So both my R9 290X cards were running at these settings without a hitch, no problem, so when I decided to go to sleep and shut down the PC for it, I quit bfgminer, naturally BSODing the system, as expected. After the PC rebooted I shut down.

Today when I wanted to start the PC up again, it worked fine for a while, then horizontal black lines appeared, (it looks like tearing tbh.)

after about 20 seconds or so, the screen just went black and when I moved my mouse after that, the PC went into a sound loop.

So naturally I tried to figure out which GPU was causing the problem and I tested each GPU individually and both showed the same problem.

The PC works fine however, when I put in my GTX 460.

The only things I could think of, but haven't tried yet, are testing the GPUs in a different PC or flashing / changing the BIOS. (The BIOS thing should be easy, as I can switch between quiet mode and uber mode, which are essentially different BIOS')

If anyone has any theories about this or knows something I don't, I would appreciate the help.

Edit: The black screen mode is NOT the screen standby mode, its displayed black, my monitor stays on while showing this black screen.

Regards
d3rrial

 

[barnsley]

What about the VRAM temps? They get very, very warm so you could've fried you VRAM. For the record as well, unless you're mining at 100% gpu fan speed they aren't covered by Warranty from what I've heard so you might be in a bit of a hole :l.

 

[d3rrial]

Do the 290X' even have VRAM temps sensors? I haven't seen those, so I can't tell you how my VRAM temps were...

 

[barnsley]

You should be able to see them on GPU from what I understand (looks like this ). Also I mean Vrm not Vram. I'm crap with names n stuff.

 

[d3rrial]

Might be possible. The strange thing is that both of my cards don't work. They worked fine the day before, and since yesterday they both appear to have the same defect, thats what stopped me from just RMA-ing them right away, hoping that desaster may be averted.

I did not check the VRM temps, will make sure to do that in the future tho >_<

An event, which might be related, which I thought was unrelated is that shortly after turning off the PC I heard the sound of something popping. (Imagine the sound of a capacitor going pop). I doubt this is related to this, but I am still mentioning it for the sake of completeness

 

[barnsley]

Might be possible. The strange thing is that both of my cards don't work. They worked fine the day before, and since yesterday they both appear to have the same defect, thats what stopped me from just RMA-ing them right away, hoping that desaster may be averted.

I did not check the VRM temps, will make sure to do that in the future tho >_<

An event, which might be related, which I thought was unrelated is that shortly after turning off the PC I heard the sound of something popping. (Imagine the sound of a capacitor going pop). I doubt this is related to this, but I am still mentioning it for the sake of completeness

it sounds like your PSU could be going maybe? The faulty psu could not be giving the 'cards the right amount of grunt under load, causing the artifacting. Got a spare psu to test one/two of them.

 

[d3rrial]

I got it to work for a while, after I reinstalled the driver, but now I have another issue: Always when I have either card installed and as graphics output device, BIOS works fine, bootloader works fine, but as soon as there should be a loginscreen, there is actually just black, I can move the mouse and the network icon in the bottom left is blinking (win 8.1) but I can do nothing else and the mousepointer resets to center every 5 seconds.

When I boot with the haswell integrated gpu tho, everything works great and I am typing this like this right now. I don't know whats going on. Please help me.

Edit: It shouldn't be a driver problem: I deleted the drivers when I was in with my haswell GPU, restarted and still had the same problem.

I have also tried the BIOS switch on the cards, no positive results

nvm.

I had a corrupted atikmdag.sys
deleted it and system runs fine now.

 

[Meladath]

Vertical lines is a common problem with AMD cards if you have unofficial overclocking enabled, used to happen to my 6950. In afterburner make sure to check "reset display mode on applying unofficial overclock" that will fix it.

If you don't have unofficial overclocking mode enabled then I am not sure sorry.

 

[d3rrial]

Well, I was overclocking unofficially. I didn't use Afterburner tho.

I will look into this, thank you very much.

 

[Surfie]

Ok, stupid troubleshooting question, but have you tested the cards independently of each other (ie. one at a time)? If so, what was the result?

 

[d3rrial]

It seems to have been a driver issue, or rather a case of GPU not liking to be overclocked in the slightest. Both cards are running again, but I can't do even the slightest overclock without them doing nasty visual glitches again.

I did test both individually and both displayed the same result, probably because of the drivers.

 

[NeverBackDown]

Any files corrupted or missing from the amd driver package or the mining software? Might just be a simple software issue that is causing this.

PSU is ruled out on the power side of things, 1000watts for 290x in crossfire is plenty.

 

[d3rrial]

Yup, I have to frequently replace the atikmdag.sys file, around every fifth reboot, this file causes a boot-loop until I replace it with a backup I have of that file, or reinstall the drivers.

 

[NeverBackDown]

found this site that lists 10 ways to fix it. Read through it all and it explains that the atimkdag.sys bsod is not a specific issue. Its a generic issue telling you something went wrong. Hopefully it is of some use to you!
Link

 

[laptopfreek0]

Same issue

Would like to chime in that I had the exact same problem and disabling overclocking on the cards fixed it as well. But honestly why do I have to disable overclocking to not get the visual artifacts. Odd if anybody finds a solution that allows for overclocking and no artifacting please let us know.

*Edit also noticed that when crossfire is enabled I can re-enable overclocking without the artifacting. Odd.

 

[NeverBackDown]

Reason why it gives artifacts is because its not stable or a driver issue is causing instability. Usually the hardware is at fault.

I provided a link above and maybe that will help solve this problem as well.