amdkmdap stopped responding

Discussion in 'Videocards - AMD Radeon Drivers Section' started by Caenyss, Nov 12, 2017.

  1. Caenyss

    Caenyss Active Member

    Messages:
    91
    Likes Received:
    28
    GPU:
    7900 XT
    Hey,

    I really need some help figuring what's happening with my system.
    I got my new parts for the 8600k system (Mobo, CPU and RAM, you can check my system on my profile I think), and ever since my display drivers have been crashing frequently only when I'm on Windows or browsing either Edge or any other browser, when gaming it hasn't happen yet.

    Screen freezes, goes black, never recovers even though the message I get from Event log is "Display driver amdkmdap stopped responding and has successfully recovered." Sometimes it gives a bluescreen refering amdkmdap.

    What I've done so far:
    - All drivers are up-to-date including the ones for this motherboard BIOS, chipset, LAN, Intel Management and so on. Windows 10 Home is up-to-date.
    - Formatted my pc for 3 times already, and I've noticed the system starts to freeze as soon as I install my GPU drivers.
    - I've tried using DDU and installing older drivers - NOK
    - I've tried using Windows Update drivers - NOK
    - I've tried using DDU and not installing any drivers - OK, but I obviously can't do much
    - System is stable, ran a few passes on Memtest (I'm not running any OC both GPU and CPU, just set XMP profile in BIOS) and I've ran a couple hours of AIDA64 stress test.
    - Current drivers are 17.11.1

    I'm really out of ideas and would greatly appreciate some help..

    Many thanks in advance.
     
  2. LocoDiceGR

    LocoDiceGR Ancient Guru

    Messages:
    2,855
    Likes Received:
    1,067
    GPU:
    Gigabyte 3060 Ti
    Power Supply ??
     
  3. Caenyss

    Caenyss Active Member

    Messages:
    91
    Likes Received:
    28
    GPU:
    7900 XT
    It has 1 year of use and it doesn't crash when I stress test the whole pc, which achieves the highest consumption..
     
  4. A2Razor

    A2Razor Guest

    Messages:
    543
    Likes Received:
    110
    GPU:
    6800XT, XFX Merc319
    ^ Happen to have any overclock on that processor? (edit: DOH, no OC) I know, I know, you've run stability tests and pushed the system to the limit, however you're describing an issue that only occurs at low load. eg, this is probably an issue that happens on the edge of clock-state changes.

    A machine can be 100% load-stable, Prime95, IBT, LINX, etc, stable, and still lock up in a web-browser or old games. If you've done any OC, try removing the OC temporarily or bumping your input voltage slightly. Since it's a brand new system and probably just has been overclocked, stress tested, and stable settings found, this has to be asked.


    EDIT: Still, despite lack of an OC, try a slight voltage bump or disabling C-States.

    EDIT #2: -- Have you tried something like Prime95 + Unigine "concurrently" in your burn testing?

    This will "probably" pass I expect ... yet if it's an instant lockup, the PSU is to blame. I would personally pin my focus on the CPU / board stability (namely in disabling all forms of processor power management and link-state power-management for PCI-E). If Prime95 passes, try LinX concurrently too (eg, LinX + Unigine -- something that will stop-go cause clock ramp-ups).

    * Processors on the edge of stable I find to do exactly what you're describing.
     
    Last edited: Nov 13, 2017

  5. Caenyss

    Caenyss Active Member

    Messages:
    91
    Likes Received:
    28
    GPU:
    7900 XT
    That was also my suspicion, regarding lower power states, but I haven't tested with C-states disabled which I am doing now.

    I still don't suspect being a PSU problem, the locks ups aren't instantly, the image freezes, my mouse cursor sometimes still moves, then stops completely, sounds sometimes keeps playing normally and then screen goes black and never recovers.

    After my initial tests with C-states/speedstep/link-state for PCI-E off, if it still happens, I'll try to slighty increase vcore and test again.

    Thank you again for you help!
     
  6. Caenyss

    Caenyss Active Member

    Messages:
    91
    Likes Received:
    28
    GPU:
    7900 XT
    So I disabled all power saving options/states and it still crashed after 10/15 minutes.

    This time I increased vcore to 1.25 while still maintaining power saving states off and it has been fine for the past half hour, which is unusual.. I'll test it some more and if this solves the issue could I have a bad CPU? Or an unstable BIOS version?
     
  7. A2Razor

    A2Razor Guest

    Messages:
    543
    Likes Received:
    110
    GPU:
    6800XT, XFX Merc319
    It could be the bios, but in my experience with these things it's usually:
    -PSU
    -motherboard
    -processor


    (*in that order, and given we've ruled out the videocard as you had no problems in the prior system it was in*)

    On the powersupply side --high ripple can cause crashes during clock changes where the same chip passes testing elsewhere.. However usually ripple gets much worse when a machine is pushed (over 70% of the PSU's capacity), hence the Unigine + CPU stress concurrently being a good idea here (to cause crashes per testing it truly being stable). Bad motherboards / bad voltage regulators, same deal. The processor might've passed its testing by Intel for example with their power delivery, yet it may need slightly higher voltage on yours. (in all cases if a small bump works, you're probably right-on-the-edge)

    bios updates 'can' effect power delivery, but that's more for their special features like load-line calibration. If your board has settings for power "filtering" you might also want to try playing with these to see if they make a difference without raising voltage as much.


    What you do is really up to you if you decide to RMA everything, you may or may not get a better chip or board (and yet again you might). Stability is pretty much subjective (hard to get absolute 100% stability) and there's alot involved here.

    ^ For example: My Xeon E5 needs 10mv added in order to be completely stable in MPC-HC with SVP 4, without I'll get crashes randomly during playback of animes (TDR reset). Yet absolutely nothing can cause the machine to fail with the exception of this (that I've found so far). Ironically, CPU is the culprit and not the videocard far as I can tell.

    With stock voltages, that same chip is ~72 hour Prime95, IBT, LinX, stable, plays all games without issues, and yet fails with MPC-HC + SVP on two different mobos without a voltage bump. (ONLY with the Fury-X, not with my older Geforce 980 which needs no change)


    --So really if it runs fine with a voltage bump, you just have to ask if you're ok with it requiring more tinkering than out of the box settings.
     
    Last edited: Nov 13, 2017
  8. Caenyss

    Caenyss Active Member

    Messages:
    91
    Likes Received:
    28
    GPU:
    7900 XT
    Thank you a lot for your input and help.

    It does seem that there are no equal parts, which can be a PITA, as it is with my case.

    I think I've finally figured out how to maintain my system stable, at least so far. I've reset my BIOS settings all to default and set for XMP, as it was before, and set CPU voltage to offset mode with an increment of 0.030V, which seems to be enough for now.
     
  9. primetime^

    primetime^ Master Guru

    Messages:
    909
    Likes Received:
    17
    GPU:
    Sapphire RX Vega 64
    try going back to 17.7.2.....it was the last stable driver for me! I was getting those errors like crazy with the newer drivers! there just full of bugs galore! i havent yet tried 17.11.2 yet so it possible they got it sorted..but either way im playing wolf2 everything maxed and enabled with zero issues. So if 17.11.2 doesn't work just go back to 17.7.2
     
  10. The Goose

    The Goose Ancient Guru

    Messages:
    3,057
    Likes Received:
    375
    GPU:
    MSI Rtx3080 SuprimX
    What memory are you running and at what speed are you running it, are you using xmp setting, the 8600k is rated for ddr4 2600hz, finally what mobo did you get.
     

  11. primetime^

    primetime^ Master Guru

    Messages:
    909
    Likes Received:
    17
    GPU:
    Sapphire RX Vega 64
    For me i can confirm a clean install of 17.11.2 should fix your issues.....At least it did on mine. I dont think the guys bios settings have anything do do with "amdkmdap stopped responding"
     
  12. A2Razor

    A2Razor Guest

    Messages:
    543
    Likes Received:
    110
    GPU:
    6800XT, XFX Merc319
    However, in this case the OP is posting about upgrading a machine (Mobo, CPU and RAM), where the videocard is presumably carried over from a known working setup (was stable before on the latest drivers). -- He's also tried a clean format of the machine multiple times with no success.

    Overall system stability has alot to do with driver crashes, odd as it may sound. I've witnessed this type of stuff many many times over the years.
     
  13. Grimbarian

    Grimbarian Ancient Guru

    Messages:
    2,106
    Likes Received:
    616
    GPU:
    RTX 3070 Ti
    Someone on the Ubi forums is having the same issue trying to play AC:O, are you the same guy? (if not if you find a fix I'll post it there!).
     
  14. Caenyss

    Caenyss Active Member

    Messages:
    91
    Likes Received:
    28
    GPU:
    7900 XT
    Hey guys,

    Sorry for not giving any updates. I managed to fix it and it was indeed a RAM/CPU/MOBO instability issue which caused all kinds of weird events, including the one I started this thread about.

    It seems that when I set XMP in BIOS my mobo, for some reason, sets certain voltages WAY too high; in this case VCCIO, System Agent Voltage, PCH Core Voltage and CPU Standby Voltage.

    When I lowered those voltages to the standard everything got stable and I haven't had any crashes for a few days now!
     
    A2Razor likes this.
  15. Caenyss

    Caenyss Active Member

    Messages:
    91
    Likes Received:
    28
    GPU:
    7900 XT
    I'm not the same guy as that one but from what I quickly gathered it doesn't seem to be the same problem as me as mine was specific with idler functions and not actual gaming.
     

  16. A2Razor

    A2Razor Guest

    Messages:
    543
    Likes Received:
    110
    GPU:
    6800XT, XFX Merc319
    Nice-- good job figuring that one out, and yep that'd do it. Higher voltages on memory directly puts more draw on the processor since all these modern chips have embedded controllers, processor voltage often has to be added in compensation. Yet if the RAM doesn't need that voltage, and you're not OCing, definitely a good thing to get that lowered.

    A proper fix is always better than a symptoms band-aid.
     
  17. Grimbarian

    Grimbarian Ancient Guru

    Messages:
    2,106
    Likes Received:
    616
    GPU:
    RTX 3070 Ti
    Ok thanks for confirming that :)
     

Share This Page