all 16 comments

sorted by: hot top controversial new old
[–] [S] 51 points 1 day ago (1 child)

At this point I've got to really wonder if the AI/memory cartels are actively sabotaging our machines to drive up prices.

  • source
  • hideshow 1 child comment
  • [–] 33 points 1 day ago* (2 children)

    If anybody is wondering, the article says it can happen with driver version 621.14 to 621.24.

  • source
  • hideshow 2 child comments
  • [–] 27 points 1 day ago

    It sounds like both companies use the same AI model to update their drivers. 🤣

  • source
  • [–] 17 points 1 day ago (2 children)

    If the billy goats can't physically get to the hardware we have at home, they'll throw "bugs" at us to kill it remotely.

    Half of us (or more) can't afford to rebuy at their artificially inflated prices. So, we'll get Chromebooks and cloud services.

    Eventually I figure we will be down to running various steam-link-like apps on devices that have just enough power to render video, connect to whatever version of the internet we are left with, and Bluetooth.

    Cunts.

  • source
  • hideshow 2 child comments
  • [–] 7 points 1 day ago (1 child)

    My thoughts as well. "Ooops we accidentally pushed a driver that killed your SSD, our bad!"

  • source
  • parent
  • hideshow 1 child comment
  • [–] 2 points 1 day ago*

    If I remember right, that quite literally happened.

    Some specific M.2 drives that had a specific type of module SkHynix or Samsung of whatever revision were affected.

    I think the drive didn't outright die but it wouldn't boot from it anymore? Something like that? It wasn't catastrophic to a power user but to a normal person it definitely was.

    It was about 1-2 years ago I think.

  • source
  • parent
  • [–] 8 points 1 day ago

    30 GB per hour is quite a bit, damn

  • source
  • [–] 10 points 1 day ago (1 child)

    A non-disk driver running disk operations doesn't sound like a good idea

  • source
  • hideshow 1 child comment
  • [–] 5 points 1 day ago (1 child)

    Wonder if AMD and NVIDIA using Claude is why both encountered the same issues not too far apart on Windows with things increasingly relying on the same source for advice. Which is also why both seem to be fine in the writing department on Linux due to Claude or whatever AI agent used didn't suggest things that would lead to issues.

    I wonder if this could be exploited to mass infect corporations as code stops being written and more and more trust is put into having AI do everything. And even QA is probably being trusted to AI.

  • source
  • hideshow 1 child comment
  • [–] 1 point 1 day ago

    this one for amd wasn't introduced by a code change, rather than a packaging change which wasn't properly evaluated after.

    I don't think there has ever been an internal test case monitoring disk r/w. I hope one was made in light of all of this

  • source
  • parent
  • [–] 1 point 1 day ago

    Good thing I haven't updated my drivers in forever

  • source
  • [–] 1 point 1 day ago

    I guess 610.88 will have to hack it for a while longer. Glad I didn't update.

  • source