Mainboard or GPU dying? Or both?

Discussion in 'General Hardware' started by lordcarlos, Jun 26, 2016.

  1. lordcarlos

    lordcarlos Guest

    Messages:
    9
    Likes Received:
    0
    GPU:
    R9 Fury
    Hi guys

    For month now I get different ~weird~ computer problems and I can't narrow it down.

    TL;DR: Top PCI-E slot does not work with Fury, but with older card. After effects + m.2 SSD = BSOD ( "Surprise Down"), Display artefacts after yt fullscreen.

    Intro: It all started with me upgrading my AMD 280x to a Sapphire Fury. That resulted in screen going black at random times, after month of bug hunting and RMA, and buying a XFX Fury it turned out to be the famous Fiji core clock bug*, that just manifested itself with screens going back instead weird glitches.

    AMD fixed that.

    Top PCI-E Lane: But by that time I already pulled and plugged the Fury card in and out of the PCI slot multiple times. At one point, after I restarted the screen would not turn on. Not even during bios start. But it would work fine in the other slots, but with 8x PCI-E speed only. Older cards like AMD R5 230 work in top PCI-E slot.

    m.2 SSD: Might be :adobe: but I will just mention it, because it might be PCI related. Sometimes, when I run After Effects and use my 950pro m2. SSD disk as cache I will get a bluescreen. Digging deep into the kernel dump:
    Code:
    !errrec ffffe0003dc4a8d8
    [....]
    ===============================================================================
    Section 0     : PCI Express
    -------------------------------------------------------------------------------
    Device Id     :
      VenId:DevId : 8086:2f04
      Class code  : 030400
      Function No : 0x00
      Device No   : 0x02
    [....]
    AER Information @ ffffe0003dc4aa58
      Uncorrectable Error Status    : 00004020 ur ecrc mtlp rof uc ca CTO fcp ptlp SD dlp und
    8086:2f04 is the PCI root port 2
    And SD bit is set / active. Apparently this means "Surprise Down", a loss in connection between the PCI-E hub and the PCI-E device.

    AMD Furi being weird: For example HDMI sound stutter. I have to turn Power Efficiency in crimson on.
    Now the .. weirdest thing ever. If I turn on power efficiency and play games like Overwatch where the core clock is all over the place, my mouse and keyboard stop working for a couple of seconds. I **** you not. It's like my last command is getting repeated over and over again. The keyboard still works, as in I can send other commands. For example I press A to walk left in a game, the problem randomly happens and I continue to walk left without pressing any thing. When I press D to walk right I just stop because the game thinks I'm pressing A and D at the same time.
    It's completely gone when Power Efficiency is off. I would not believe it if it was not myself it happens to.

    Display artefacts after exiting youtube fullscreen: Just happend in the last week. Not every time. Goes away again when I reset the driver. Video:
    Code:
    youtube.com/watch?v=5P1-zj5wYt8
    looks very similar to the famous Fiji core clock bug, but that should have been fixed.

    Ram: Multiple hours of memtest did not show any errors!

    Question: I hate my fury and will RMA it, but should I also RMA my Mainboard?

    Thanks for your time reading this wall of text.

    *Sorry, I can't post links yet, search for: amd "Fury X Display problem."

    System:
    OS: Win 10
    CPU: Intel Core i7-5820K
    Mainboard: MSI X99S SLI PLUS (MS-7885)
    Ram: 32gb DDR4
     
  2. lordcarlos

    lordcarlos Guest

    Messages:
    9
    Likes Received:
    0
    GPU:
    R9 Fury
    This all happend over the mast few month. (from ~December 2015)
    And I already sold my 280x.

    I will probably send my XFX Fury back and buy a 480, but if there is something wrong with the motherboard I'm scared it will fry the new card :S
     
  3. thatguy91

    thatguy91 Guest

    A BIOS update could help with a possible incompatibly.
     
  4. jura11

    jura11 Guest

    Messages:
    2,640
    Likes Received:
    707
    GPU:
    RTX 3090 NvLink
    Hi there

    I don't this is issue of the GPU if you will bit of search then you can find bit more information on that

    Please have look on this thread over on overclock.net

    http://www.overclock.net/t/1539708/...ou-see-pcie-bus-errors-please-respond-to-poll

    Looks like this is issue of the X99 on W10,I would try install W7 or W8 if you have spare HDD,I'm still on W7 and I'm not looking to upgrade anytime soon,plus I do have installed Yosemite and everything works for me

    Hope this helps

    Thanks,Jura
     

  5. thatguy91

    thatguy91 Guest

    Probably resolved with a bios update :). Isn't the PCI-E lanes controlled through the CPU? If there is errata a CPU microcode update would possibly help? Bios update should be up to date with this, or you can use UEFI Bios Updater to update generic modules before flashing.
     
  6. lordcarlos

    lordcarlos Guest

    Messages:
    9
    Likes Received:
    0
    GPU:
    R9 Fury
    Thanks, updated bios, though the change log did not say anything about bug fixes. Just support for new memory and CPUs. Lets assume the BSoD is a known issue and of my control for now.

    Leaves me with a card that does not work in the top PCI slot. + weird keyboard mouse behavior.

    Should I just assume the card is weird and RMA it?
     
    Last edited: Jun 29, 2016
  7. I might have a solution to your problem with the keyboard + mouse behavior because I had a similar issue when I first used Win10, although the difference being my mouse would not respond at Windows login (but would at boot) until I plugged it back in the usb port.

    I solved my keyboard + mouse losing power at boot by going to the device manager (right-click start button -> device manager) then doing the following:

    - In the device tree find the universal serial bus controllers sub-tree and expand the tree of usb controllers

    - Go to the first device of many named USB Root Hub, right-click device and select properties

    - Select the Advanced Power Management tab, then uncheck the option "Allow the computer to turn off this device to save power"

    - Repeat the last two steps for all devices named USB Root Hub


    As for the GPU issue, first try checking the PCI-e settings in the bios for the top PCI-e x16 slot is set to receive the maximum 75 watts allowed through the slot. Also, you might try cleaning the gold contacts on PCI-e connector on the graphics card with 91%+ isopropyl alcohol (rubbing alcohol) applied to a q-tip or lint-free/microfiber cloth, then let alcohol evaporate before putting card back in PCI-e slot.
     
  8. lordcarlos

    lordcarlos Guest

    Messages:
    9
    Likes Received:
    0
    GPU:
    R9 Fury
    Already did that some time ago, does not help.

    I might look into that.
    Just today I needed to restart again just because I could not watch a video without audio interferences. Changing the Power Efficency toggle sometimes helps, but the amount of debugging time I put into this card is just not worth it. Time for a 1070.
    My 280x did so well :)

    Thanks all!
     
  9. lordcarlos

    lordcarlos Guest

    Messages:
    9
    Likes Received:
    0
    GPU:
    R9 Fury
    Got money back for my Fury \o/
    1070 is working great so far. No mouse/keyboard problem, Top 16x PCI slot working, no glitches :)
     

Share This Page