Radeon Graphics cards Black Screen bug is haunting AMD

Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Feb 14, 2020 at 9:52 AM.

  1. Hilbert Hagedoorn

    Hilbert Hagedoorn Don Vito Corleone Staff Member

    Messages:
    37,002
    Likes Received:
    6,075
    GPU:
    AMD | NVIDIA
    chispy and Embra like this.
  2. FrostNixon

    FrostNixon Master Guru

    Messages:
    209
    Likes Received:
    28
    GPU:
    RX 5700 XT
    Main reason why this is my last AMD card. 3060 can't come any sooner.
     
  3. andries

    andries Active Member

    Messages:
    89
    Likes Received:
    3
    GPU:
    Galax 2080 ti
    Only Navi!? I had the exact same issue on Hawaii, such denial around this issue...
     
    Alessio1989 and prazola like this.
  4. Undying

    Undying Ancient Guru

    Messages:
    12,486
    Likes Received:
    1,932
    GPU:
    Aorus RX580 XTR 8GB
    This is gone far enough, as soon drivers improve little bit they add another list of brekaing issues.
    Amd should just keep making cpus and let nvidia nvidia and intel make gpus. Its sad the state radeon group is right now. A big Navi? Who the hell cares.
     

  5. Andy Watson

    Andy Watson Member Guru

    Messages:
    131
    Likes Received:
    31
    GPU:
    960
    A good read and agree with the points raised.

    It's such a shame the software side is putting a black mark against finally some competitive hardware on the graphcs side. They need to spend more on the driver team and QA / bug fix team.
     
  6. prazola

    prazola Member Guru

    Messages:
    172
    Likes Received:
    14
    GPU:
    R9390XSOC / R9290DCU2OC
    Same, and I think it's related to the hardware acceleration, but only on windows. I've never managed to trigger the black screen in my linux systems.
     
    OnnA, Silva, andries and 1 other person like this.
  7. JonasBeckman

    JonasBeckman Ancient Guru

    Messages:
    16,045
    Likes Received:
    1,943
    GPU:
    AMD S. 5700XT Pulse
    It's various things though hardware acceleration is kinda botched too as is some issue around multiple-displays and what else is actually causing this display signal loss up to a reboot being needed though at least this should be the last remaining of the black screen problems although I am curious how many actual underlying problems there are and what causes them.

    Well fixing that up so the GPU doesn't go blergh every time the driver goes whoops would be nice, not having some of these issues around late 2019 and particularly the early Adrenaline 2020 drivers regress and affect Vega and Polaris (And more?) would have been good and someone has to re-fix that Radeon VII and it's various problems as that's kinda still AMD's sorta flagship GPU so yeah that all needs works which well it's unfortunate how this has developed and the time it's taken, good hardware for Navi and all but the way the driver issues kinda rolled down and affected GCN and left existing issues open while these had to (And still have to.) be fixed is quite a problem.

    And yeah it seems the current Linux 5.5 kernel and upcoming 5.6 core and AMD's code contributions together with the changes to both their driver and the open source variant are doing better than the current Windows drivers, maybe take some of those code fixes and get them into the Windows code ha ha. (Eh it's never quite so easy.)


    EDIT: Would be nice with a easy way to replicate it though but it's kinda random so aside from a few more sensitive games like Battlefield 5, Witcher 3, Borderlands 3 I think come up frequent enough but then these are also popular big sellers so more users would report on these titles anyways the randomness also makes it kinda hard to do anything about and fixing one thing like the system stability with hardware acceleration could still see a black-screen error in a game at whatever point.

    Lots of reading and some things help but still no real solution or idea what is causing it.
    Well the driver crashes so it's something with that but then why is it crashing and how and if it's hardware or software and overall compatibility concerns and what else. (Bit of everything's been tested it seems but still nothing conclusive.)


    I wouldn't really know though, just been doing a lot of reading up on it and as a Navi user it's gotten better but the issue is still not completely fixed and if there are changes in hardware it doesn't seem the 5500 and 5600 have done anything to resolve it as it happens on users with these GPU's too.
    (Newer GPU's and less reports from these as a result but it might grow a bit too depending on if the upcoming drivers will resolve this or if it'll be a pending issue for who knows how long.)
     
    Last edited: Feb 14, 2020 at 10:55 AM
  8. Mitch 74

    Mitch 74 New Member

    Messages:
    7
    Likes Received:
    5
    GPU:
    RX480 8Gb
    There are bugs. And there are Schroedinbugs. Looks like one to me.
     
  9. Astyanax

    Astyanax Ancient Guru

    Messages:
    4,734
    Likes Received:
    1,403
    GPU:
    GTX 1080ti
    on windows the browser acceleration and applications based on CEF are going to be DMA heavy, since we know NAVI has SDMA faults this can all correlate back to a stepping issue that will be nigh impossible to fix in drivers.
     
  10. JonasBeckman

    JonasBeckman Ancient Guru

    Messages:
    16,045
    Likes Received:
    1,943
    GPU:
    AMD S. 5700XT Pulse
    Hmm wasn't SDMA mentioned as one of the fixes for the 5500 which should also be on the 5600 being more recent still and thus it should have those in hardware too (Rumors around a newer stepping and all that.) though I suppose we will see how these differ from the 5700 once AMD resolves this yet again in a newer driver release (And maybe this time don't break it in another driver afterwards.) and in particular it looks like the 80.x builds of Chromium are more sensitive so that would make a good stability test.

    EDIT: Or perhaps the SDMA fixes on the 5500 were just rumors, need to look that up then.

    EDIT: Well there's something in the Linux drivers at least working around it by disabling it but that report was for Polaris though filed in January 2020

    https://www.phoronix.com/scan.php?page=news_item&px=RadeonSI-Disables-Polaris-SDMA

    Windows ones being closed source makes it tough to look into changes like this but the Linux driver situation is a bit more open and can give some info on occasion even if it doesn't apply to Windows directly.
     
    Last edited: Feb 14, 2020 at 11:03 AM

  11. s88urd

    s88urd Master Guru

    Messages:
    362
    Likes Received:
    4
    GPU:
    Vega 56
    I would like to add the following.
    If using a PCI-E extension or any sort of vertical GPU riser on an X570 Board, ensure your BIOS is set to PCI-E Gen 3.

    I was facing lots of weird issues after I installed a vertical riser with my 5700XT (on an x570 board) and it turns out the majority of these riser kits do not support PCI-E Gen 4.
    Within the BIOS i forced it to use PCI-E Gen 3 and this resolved the issue for me.
     
    P.O.N. and Silva like this.
  12. Kaarme

    Kaarme Ancient Guru

    Messages:
    1,781
    Likes Received:
    521
    GPU:
    Sapphire 390
    You must be a major Intel stock owner if you enjoyed that much the drought in the x86 CPU world for a decade. Now you are hoping for the same on the GPU side? Give me a break.

    All people should hope for is that AMD fixes their issues, not that they would exit the market. Unless you are a major Nvidia stock owner and only care about the dividends, not about the technology development and product availability.
     
  13. butjer1010

    butjer1010 Member

    Messages:
    44
    Likes Received:
    8
    GPU:
    Asus RX5700 XT
    I don't have any of these problems on mine...... Asus RX5700 XT???? Works excellent.....
     
    rflair, airbud7, OnnA and 1 other person like this.
  14. andries

    andries Active Member

    Messages:
    89
    Likes Received:
    3
    GPU:
    Galax 2080 ti
    What technology development are we talking about here? Real time ray tracing perhaps? Or the fact that Amd has not been able to rectify the black screen issue since the start of a decade?
     
  15. SpajdrEX

    SpajdrEX AMD Vanguard

    Messages:
    2,188
    Likes Received:
    499
    GPU:
    Sapphire RX5700XT
    I know it's hard for you guys to wait for fixes, but we are working hard to reproduce and fix the bugs as soon as possible.
     

  16. Undying

    Undying Ancient Guru

    Messages:
    12,486
    Likes Received:
    1,932
    GPU:
    Aorus RX580 XTR 8GB
    Who is working? No one tells us anything. Didnt you had an nvidia card two weeks ago and now your are part of amd driver team?
     
  17. SpajdrEX

    SpajdrEX AMD Vanguard

    Messages:
    2,188
    Likes Received:
    499
    GPU:
    Sapphire RX5700XT
    I'm part of AMD Vanguard beta team. NVIDIA card I had for testing specific games (which use AnvilNext 2.0 engine), as they have issues on AMD cards, at least with slower CPU's.
    This was not required to be tested, but I wanted to show some results and then sold the card.
     
    Undying likes this.
  18. JonasBeckman

    JonasBeckman Ancient Guru

    Messages:
    16,045
    Likes Received:
    1,943
    GPU:
    AMD S. 5700XT Pulse
    The Vanguard team is under some pretty strict NDA's and I would expect they aren't entirely involved in the internal discussions and developments for AMD even if there is some AMD staff involved in handling this.
    (Well they would be informed to some degree but still a external group of testers even with said NDA signed and what that allows and prevents talking about for ongoing developments and current issues and fixes in progress.)

    I never opted in to it but there's probably a fair amount of users but due to the nature of this black screen problem being so hard to replicate easily and how random it is overall even with some hundred involved Vanguard testers or what the total group size might be it's still possible a large majority of these are unable to actually replicate several of the reported problems making testing very difficult even if using known titles that can be prone to stability issues.

    Interestingly there's also the Gamer Nexus report where they took home a few Navi's and hit a variety of issues unable to be replicated in the test environment and benchmarks normally and then there's also the recent Hardware Unbox report where the guy had a few Navi GPU's at home but they all worked fine so yeah there's just such a large amount of different factors and possibilities for what is causing this and it's tough to get it replicated and reported in enough detail to where it can be tracked down and resolved I would imagine as one of the key reasons behind the time it's taken though I don't know why AMD was so late with acknowledging the problem in the first place unless there was some internal discussions around what could be said or a large chance of user errors and hardware compatibility causing this so not directly related to AMD's hardware or the drivers.


    EDIT: And then for Vega and Polaris that should be purely in software as those were working fine before but then that also opens up for it being a multitude of different issues with similar effect which for whatever reason kinda shuts down the GPU instead of a recoverable crash on the software side or a BSOD if it's going into Kernel mode somehow.
    (Think a large part of the driver model is in user mode so it can happen but should be rare short of something going very wrong or being hardware related.)
     
    Keitosha, Maddness, airbud7 and 2 others like this.
  19. SpajdrEX

    SpajdrEX AMD Vanguard

    Messages:
    2,188
    Likes Received:
    499
    GPU:
    Sapphire RX5700XT
    Hey Jonas, you should try to join us, when I have free time I'm always reading your wall of texts :)
     
    Maddness and OnnA like this.
  20. rl66

    rl66 Ancient Guru

    Messages:
    2,286
    Likes Received:
    163
    GPU:
    quadro K6000+Tesla M2090
    On other hand i had artefact and graphic issue with nearly every generation of NVidia gen since fermi using a non windowed game + in it's time TeamSpeak then Discord...
    Ok artefacts and texture problem are not blank screen, BUT it is issue too...
    It is clearly a windows related problem as someone already said, on other OSes it doesn't appear.
     

Share This Page