• HStone32@lemmy.world
    link
    fedilink
    arrow-up
    7
    ·
    1 hour ago

    Its not just Linux. NVIDIA sucks on windows too, but those PCMR types aren’t willing to admit it.

  • Zacryon@feddit.org
    link
    fedilink
    arrow-up
    27
    ·
    5 hours ago

    It’s really ironic and embarassing. The most valuable chip manufacturer in the world, thanks to advances in AI and AI research, which is usually done using Linux systems. And yet Nvidia still sucks hard when it comes to Linux support.

    • brucethemoose@lemmy.world
      link
      fedilink
      arrow-up
      20
      arrow-down
      1
      ·
      edit-2
      3 hours ago

      Nvidia graphics support.

      Their money is in headless systems, which TBF are much less problematic with Nvidia. Anything CUDA is first class on Linux with Windows as an afterthought.

      • Zacryon@feddit.org
        link
        fedilink
        arrow-up
        4
        ·
        3 hours ago

        Anything CUDA is first class on Linux with Windows as an afterthought

        Not in my experience.

    • Jesus_666@lemmy.world
      link
      fedilink
      arrow-up
      1
      ·
      1 hour ago

      I wouldn’t call their Windows support stellar, either. There’s only one error code for any and all problems and RTXes can be damn finicky if you’re unlucky.

    • Petter1@lemm.ee
      link
      fedilink
      arrow-up
      1
      ·
      1 hour ago

      Latest kernel, wayland, NVIDIA dGPU greater than RTX 900, latest proprietary NVIDIA driver and correct vulkan packages for said driver:

      Don’t see any issues, really

      What does not work?

      Maybe some install steam and choose wrong vulkan dependency 🤔

      Or have vulkan dependency for nouveau installed prior to switching to nvidia drivers

      I just wonder why pacman not automatically chooses the correct dependency 😅 but I am sure there is a reason

  • traches@sh.itjust.works
    link
    fedilink
    English
    arrow-up
    7
    arrow-down
    1
    ·
    4 hours ago

    Yeah, I don’t buy nvidia for this exact reason. No amount of performance matters if the drivers are broken

  • Shortstack@reddthat.com
    link
    fedilink
    arrow-up
    37
    ·
    7 hours ago

    This is why I don’t update Nvidia drivers unless there’s a fix I need. If it’s stable, that’s all I want.

    • NatanoxOP
      link
      fedilink
      English
      arrow-up
      18
      ·
      5 hours ago

      Don’t worry, Nvidia got you covered: it also likes to occasionally break with kernel updates.

    • sanpo@sopuli.xyz
      link
      fedilink
      arrow-up
      29
      ·
      6 hours ago

      Until the GPU cooks itself anyway, because nvidia can’t admit their new power connector was a mistake.

      • Schmuppes@lemmy.today
        link
        fedilink
        arrow-up
        17
        arrow-down
        1
        ·
        edit-2
        5 hours ago

        Seems to be less about the connector, but more about load balancing. The German guy who had 150°C connectors at the PSU side also measured current draws. One cable was doing 22 A (so almost half of the 5090’s total consumption) while the other 7 five were just chilling.

        • sanpo@sopuli.xyz
          link
          fedilink
          arrow-up
          2
          ·
          2 hours ago

          How about both?

          I’d expect the design to take into account this kind of issue, they’re only one of the most valuable companies in the world, surely they can afford some QA.

        • exu@feditown.com
          link
          fedilink
          English
          arrow-up
          7
          ·
          4 hours ago

          Nvidia connected all six cables like they were one and have no way to measure or balance the load across all six.
          They used to do load balancing on the 30 series, treating it as 3 cables basically (3x2 cables).

        • RamblingPanda@lemmynsfw.com
          link
          fedilink
          arrow-up
          6
          ·
          5 hours ago

          That’s true, but once my apartment is on fire, I don’t really care if it was the cable or the connector.

          My insurance might be interested, though.

    • Petter1@lemm.ee
      link
      fedilink
      arrow-up
      3
      ·
      1 hour ago

      Not just some…

      They should just buy a PC build for linux with it preinstalled

      Like, tuxedo, system67(or whatever number they have) or I think framework is available like that as well.

      • SmokeyDope@lemmy.world
        link
        fedilink
        English
        arrow-up
        2
        ·
        edit-2
        57 minutes ago

        Except all those Linux specific privacy/modularbility orientated PCs are expensive as hell with base models starting close to 1k last I checked. They’re aimed at the demographic of cushy tech nerds making bank tapping at a keyboard who care about opsec or right to repair and can comfortably afford dropping 1k on a new laptop without thinking twice about if that money has better uses.

        I and many others who don’t live in economic lah-lah land will NEVER be able to justify 1k spent on a laptop just because it has physical kill switches or modular parts and preloaded with a good Linux distro. These companies need to touch grass and come down a couple hundred dollars to the 400-500$ range then we can talk. Until that day comes, the guy selling librebooted thinkpads on eBay running popos or mint is the better option for those who live with the reality of not having a lot of money.

    • Tetsuo@jlai.lu
      link
      fedilink
      arrow-up
      8
      ·
      6 hours ago

      No, just upgraded from a working computer to an up to date not working computer.

    • NatanoxOP
      link
      fedilink
      English
      arrow-up
      1
      ·
      5 hours ago

      Not sure with a distro using apt. If it was zypper (i.e. OpenSuse) then I could confirm that you’re in for some trouble since Nvidia messed up their package repo for it.

      In case anyone with an OpenSuse Tumbleweed stumbles over this: in case it broke for you you have to purge any Nvidia related package, remove the repo, make sure anything Nvidia-related in your modprobe.d folders are gone (don’t forget the one in /usr/ !), run dracut -f and reboot. Then reinstall the Nvidia repo and run this command while ignoring the one “recommended” package he skips and, nost inportantly, if he asks you about what to do about the lacking dependency of openssl1.1 you want to proceed while breaking it (Nvidia fucked up dependencies completely):

      sudo zypper install libnvidia-egl-gbm1 libnvidia-egl-wayland1 libnvidia-egl-x111 nvidia-common-G06 nvidia-compute-G06 nvidia-compute-utils-G06 nvidia-driver-G06-kmp-default nvidia-gl-G06 nvidia-libXNVCtrl nvidia-modprobe nvidia-persistenced nvidia-settings nvidia-video-G06 libopenssl3

      Run dracut -f and reboot. It should work now.

  • LainTrain@lemmy.dbzer0.com
    link
    fedilink
    arrow-up
    1
    ·
    7 hours ago

    Iktf, I had my 1060-6GB die and for a while I was gaming on a 750Ti lol.

    Recently my Crucial 1TB SATA SSD suddenly died, no errors, no SMART, no detection on any computer or via USB adapter, only coil whine, taking with it exactly all the things I never backed up because SSDs are supposed to be good :(

    • tiramichu@lemm.ee
      link
      fedilink
      arrow-up
      10
      ·
      edit-2
      7 hours ago

      SSDs at the end of their lifespan do tend to fail more gracefully than HDDs, as even when they become fully worn and unable to take new writes, they will often still allow reads.

      But, that depends on the specific type of failure.

      I had an SSD fail in the same way as yours, where the controller chip or something along the path there died, and it went from fully working to toast in an instant.

      Some drives are more reliable, some drives are less reliable, but the only rule is that any drive can break, at any time, old or new.

      Always have backups.

      • LainTrain@lemmy.dbzer0.com
        link
        fedilink
        arrow-up
        2
        ·
        6 hours ago

        Yeah i should know, but I’m too lazy haha. Didn’t lose anything completely irreplaceable but my beautiful bind9 local DNS zone written and annotated by hand is gone.

        Plus I have basically nowhere to back up to.

        At least the first thing I did when reinstalling Debian was set up an an rsync cron job to fetch the home, etc and some other select dirs, but this is backing up to a Raspberry Pi with a busted micro SD slot that runs off a rather dodgy USB enclosure’d 120 gig mSATA SSD that already failed before that originally transplanted from a busted MSI gaming laptop I sold for coke cash in the mid-2010s.

        Not ideal. That pi also periodically shits the bed. It’s exposed to the elements a bit because it’s also in use in 2 DIY iot projects.

        Is there a decent non-shit non-megacorp-empowering affordable way of doing off-site backups on a small scale?

        • tiramichu@lemm.ee
          link
          fedilink
          arrow-up
          1
          ·
          6 hours ago

          Sounds like you’ve got quite the esoteric setup, hehe :)

          My personal solution isn’t exactly small as I have two identical four-disk NAS servers which operate with one as primary and the other as a read-only mirror of the primary. For off-site I don’t have an automated solution but just backup onto external every so often and leave it with a family member.

          A good solution could be as simple as a raspberry pi with an external SSD at a friend or family’s place, and then make that accessible via VPN to your home network.

          • NatanoxOP
            link
            fedilink
            English
            arrow-up
            1
            ·
            5 hours ago

            For smaller amounts you can also get away with Cryptomator and some free account at Google, Onedrive or whatever. 😉