Thread stuck in device driver error - XFX RX 480s don't work - tried everything

CJLWMiningCJLWMining Member Posts: 9
Hi Community :) ,
I ran into a Problem while building my ETH mining rig. Setup:

4 x XFX Radeon RX480 Black edition
Equipped with (as of yet) 3 x Artic Accelero Hybrid III 140mm
Asus P10S WS
Intel i3 6100
1000W BeQuiet! Dark power Pro
Kingston HyperX fury 8Gb DDR4 something or other (I believe this is not important)

While doing some initial test runs, one of the cards gave out and I received a BSOD with "Thread_stuck_in_device_driver_error". Since then I cannot get windows going, with the card plugged in and drivers installed (it works with generic Windows drivers or in safe mode). I am confused about the fact, that the card itself isn't ignored at start up, and the rest of the hardware would keep going. Even the internal GPU doesn’t work when the card is plugged in.
The Microsoft Bug Check Website about the Error message is too specific for me as I am not a programmer.

Before this, the Mboard was already showing CPU over voltage error, which I ignored since all seemed fine in CPU-Z.

I reran the miner with the three remaining GPUs. In the night (coincidentally around the same time as the other one) the PC caved and another GPU showed similar symptoms. Surprising result: This time around, the GPU with the stock fan was the problem, causing me to believe that there is something wrong with either the PSU, the Mboard, or the miner (Up to this point, I blamed overheating of one of the VRMs or VRAMs in the GPU). Another weird thing: The second broken card continued to work for about a day in my old PC. When it stopped, it still worked in the second PCIe slot, but without any display output (I could access windows through TeamViewer, but the card was not recognised as a display adapter in windows’ device manager). Now I am properly confused about the whole thing.
I tried to flash the BIOS on the first broken card to no avail. BIOS flash works like a charm, but driver installation is still not possible afterwards. Can anybody tell me which devices on the card are active when flashing BIOS as compared to installing the driver?
Also, I cannot seem to get Ubuntu going, no matter whether the (hopefully not) damaged card is plugged in or not. It just shows the desktop background without any symbols. I can get into the command line but installing a driver doesn't work either.

I've already tried a clean install via DDU by Wagnard.

All help would be greatly appreciated, since I am absolutely stumped. Thank you!

Comments

  • muzzy124muzzy124 Member Posts: 78
    edited October 2016
    1. remove all cards
    2. ddu drivers
    3. install one card and drivers
    4. check if it works, reboot
    5. install second card... etc

    in my case xfx cards are very strange, one is already RMAed with buggy ram, other three behave not so perfect as sapphires
  • CJLWMiningCJLWMining Member Posts: 9
    Thanks for your quick reply, I will try it asap. - The XFX cards looked great on paper, but ill try saphire next time...
  • cidmocidmo Member Posts: 446 ✭✭✭
    i have been having weird issues with windows device installer or whatever it is not applying the driver right
    if u can see the card in device manager right click into the properties windows
    for some reason as soon as i do that windows realizes it knows what the card is and applies the driver
    makes no sense but windows 10 seems to be a pieced together ball of crap to begin with
  • CJLWMiningCJLWMining Member Posts: 9
    muzzy124 said:

    1. remove all cards
    2. ddu drivers
    3. install one card and drivers
    4. check if it works, reboot
    5. install second card... etc

    in my case xfx cards are very strange, one is already RMAed with buggy ram, other three behave not so perfect as sapphires

    I tried DDUing without cards in place, but when installing the faulty cards it just returns the same thread stuck in device driver error, as soon as it is applying the drivers to the card (about 20 seconds after windows boot).

    Is it possible that the 16.9 ATI drivers are causing problems with the claymore miner?
  • CJLWMiningCJLWMining Member Posts: 9
    edited October 2016
    UPDATE: Weird things happening again:
    I plugged the card into (yet) another PC and the drivers installed just fine. I did run it under full power for a couple of seconds, but stopped in order to not damage the remaining PC in any way (it is not mine). But from the experience with the second PC, I would assume that it would give out at a certain point and then not work with the card anymore.

    Completely stumped now. All BIOSs involved were flashed, so there should not be any "remembering" of the cards by the Mboard. What could be the problem, when the PCIe ports are apparently fine?

    I also obviously can not send "working" cards back to XFX.

    This is gnawing at my rationality!! Please help! :s:'(
  • cvipercviper Member Posts: 132 ✭✭
    @CJLWMining
    Its obviously a GPU problem send it to XFX they will check it and tell you if its damaged or not.
    I had a problem with Watchdog stuck thread on opencl or something like that error in claymore wich will crash my PC soon but it was only present when underclocked core / overclocked ram.So i had to reinstall a fresh new windows and unplugged and plugged all cabels again,it turned out to be a slightly loosed riser caber x1 to the motherboard that was causing it.
    Try theese settings on ur system and tell us if its more stable https://forum.ethereum.org/discussion/8262/rx480-4gb-and-8gb-settings-crimson-16-7-1-win10/p1
  • CJLWMiningCJLWMining Member Posts: 9
    thanks for the reply :smile:

    I will talk to XFX and ask whether they will take in the modified card as well. Since it has happened to two cards though this isn't really a final solution for me, since I can not rerun the miner without risking another card failure.

    Weirdly, after taking the boxed card back home it started working in the rig again :/
    So something I did with the other (3rd) PC "healed" the card (once again proving, that it is not completely broken). One of them is still out of order though and I am not comfortable running the miner again. Has anyone got any other ideas?

    Thank you for your support!
  • SmokyishSmokyish Member Posts: 203 ✭✭
    So it sounds like you propably had a riser that wasn't properly seated or an issue with the riser, ie. broken one. When troubleshooting, you usually wanna start with checking the riser(s), since they often are the cause of your problems ;)
  • CJLWMiningCJLWMining Member Posts: 9
    Thanks for your reply :smile:

    But that would not explain why the card does not work in the other PCIe slots and stopped working in the second PC with the same issue. Is it likely that the cards burnt out some connection in the riser both in the first and the second PC, due to some incorrect power config?
  • cidmocidmo Member Posts: 446 ✭✭✭
    could be mobo issue as well
    i bought an amd mobo with a a6 chip, everything worked fine 1-4 cards in any pci-e slot
    actually liked the board had a bunch of new stuff like USB 3.1
    as soon as i added a 5th card in any slot with a riser or not it would corrupt the bios
    and whatever corrupted the bios would kill the backup bios as well even if u started without anything in it
    gigabyte was no help, tried this on 2 mobos and fried bios on both
    so obviously gigabyte just added stuff to an old chipset never fully testing it
    not saying ur having a similar issue but might explain the diff between 2 mobos
  • CJLWMiningCJLWMining Member Posts: 9
    You might be right. I read up on the "CPU over voltage error" that I get everytime I start the rig. It seems to be a common problem with ASUS Mainboards. I guess that might have cooked a certain part of both GPUs. Just very weird that the card kept going for a while in the other PC.

    Also small update: The Card that was "healed" is now not working anymore. Just died on me while browsing. I will send it to XFX and ask ASUS about taking the Mboard back. That should give me the confidence to rerun the miner with a stock card.

    I am still open for suggestions of course :smile: All in all this seems to be one curious issue
  • LickMyNvidiaLickMyNvidia Member Posts: 36
    I had this exact same problem and here is how I fixed it.

    1. Start from zero. Strip down the machine and reinstall Ubuntu with ONE GPU and your monitor connected to the GPU.
    2. Download and install AMDGPU-PRO Driver for Linux v16.30 Here ----> http://support.amd.com/en-us/kb-articles/Pages/AMD-Radeon-GPU-PRO-Linux-Beta-Driver–Release-Notes.aspx
    3. Shut down system.
    4. Install second GPU (in the slot recommended by the manual for multi GPU's) and reboot.
    5. Shut down system and repeat until all 4 cards are installed.

    At this point everything should be fine. If you boot the system and just get a background with nothing on it or you get a Desktop with no icons and your sidebar has turned into a drop down on the top left of your screen you have installed the cards in the wrong order and need to start over from (1). Once everything is correctly installed you can download your miner and get to work. In my case only the first GPU installed will send video out.

    Good Luck and let me know if this worked for you.
  • drizzt_do_urdendrizzt_do_urden Member Posts: 181 ✭✭✭
    I had a card with Thread stuck in device driver. Tried everything but nothing worked. Finally I sent it back to the vendor and they replaced it right away. turns out it was a faulty card from factory.
  • SomeoneHereSomeoneHere Member Posts: 4
    It's properly cause due to bad vbios flash. I did and take the same result. So i give it back to store and take new one.... Lol
  • CJLWMiningCJLWMining Member Posts: 9
    edited October 2016

    I had this exact same problem and here is how I fixed it.

    1. Start from zero. Strip down the machine and reinstall Ubuntu with ONE GPU and your monitor connected to the GPU.
    2. Download and install AMDGPU-PRO Driver for Linux v16.30 Here ----> http://support.amd.com/en-us/kb-articles/Pages/AMD-Radeon-GPU-PRO-Linux-Beta-Driver–Release-Notes.aspx
    3. Shut down system.
    4. Install second GPU (in the slot recommended by the manual for multi GPU's) and reboot.
    5. Shut down system and repeat until all 4 cards are installed.

    At this point everything should be fine. If you boot the system and just get a background with nothing on it or you get a Desktop with no icons and your sidebar has turned into a drop down on the top left of your screen you have installed the cards in the wrong order and need to start over from (1). Once everything is correctly installed you can download your miner and get to work. In my case only the first GPU installed will send video out.

    Good Luck and let me know if this worked for you.

    Thank you :smile:
    I did not have time to send the cards/mboard back yet so I'll try that out later today. Up to now, Ubuntu wasn't functioning properly but I didn't really press the issue yet, since I figured that there was a hardware problem anyway.

    It's properly cause due to bad vbios flash. I did and take the same result. So i give it back to store and take new one.... Lol


    Bad vbios flash is always an option, however flashing the vBIOS is working fine with custom BIOS builds, but does not have any effect on the cards behaviour.

    I'll keep you posted :wink:
    Post edited by CJLWMining on
  • CJLWMiningCJLWMining Member Posts: 9

    I had this exact same problem and here is how I fixed it.

    1. Start from zero. Strip down the machine and reinstall Ubuntu with ONE GPU and your monitor connected to the GPU.
    2. Download and install AMDGPU-PRO Driver for Linux v16.30 Here ----> http://support.amd.com/en-us/kb-articles/Pages/AMD-Radeon-GPU-PRO-Linux-Beta-Driver–Release-Notes.aspx
    3. Shut down system.
    4. Install second GPU (in the slot recommended by the manual for multi GPU's) and reboot.
    5. Shut down system and repeat until all 4 cards are installed.

    At this point everything should be fine. If you boot the system and just get a background with nothing on it or you get a Desktop with no icons and your sidebar has turned into a drop down on the top left of your screen you have installed the cards in the wrong order and need to start over from (1). Once everything is correctly installed you can download your miner and get to work. In my case only the first GPU installed will send video out.

    Good Luck and let me know if this worked for you.

    Thank you :smile:
    I did not have time to send the cards/mboard back yet so I'll try that out later today. Up to now, Ubuntu wasn't functioning properly but I didn't really press the issue yet, since I figured that there was a hardware problem anyway.
    Well that didn't work... With one card there is just a blank screen coming up in place of Ubuntu. With the other, the fan boosts to 100% and it's trying to load the driver with these errors:

  • NourStarNourStar Member Posts: 12
    what happen for your card what did u do ? back to vendor or fixed it ? i have the same problem and warranty is gone i hope u have a solution :smile:
Sign In or Register to comment.