• 0 Posts
  • 89 Comments
Joined 1 year ago
cake
Cake day: June 18th, 2023

help-circle







  • That seems to require a level of foresight and planning that most corporations don’t have. That’s almost like a blueprint for failure when some middle manager changes the scope of a project with a hard coded time limit, IMO.

    Anyone interested in not-agile development? Maybe we can call it “Ship it when it’s ready” lol





  • Naz@sh.itjust.workstoMemes@lemmy.mlGet rich quick
    link
    fedilink
    arrow-up
    20
    ·
    edit-2
    17 days ago

    I’m an AI Developer.

    TLDR: CUDA.

    Getting ROCM to work properly is like herding cats.

    You need a custom implementation for the specific operating system, the driver version must be locked and compatible, especially with a Workstation / WRX card, the Pro drivers are especially prone to breaking, you need the specific dependencies to be compiled for your variant of HIPBlas, or zLUDA, if that doesn’t work, you need ONNX transition graphs, but then find out PyTorch doesn’t support ONNX unless it’s 1.2.0 which breaks another dependency of X-Transformers, which then breaks because the version of HIPBlas is incompatible with that older version of Python and …

    Inhales

    And THEN MAYBE it’ll work at 85% of the speed of CUDA. If it doesn’t crash first due to an arbitrary error such as CUDA_UNIMPEMENTED_FUNCTION_HALF

    You get the picture. On Nvidia, it’s click, open, CUDA working? Yes?, done. You don’t spend 120 hours fucking around and recompiling for your specific usecase.









  • Sure, I’ll try OpenSUSE!

    Tumbleweed is a bit of a spooky name for a distro implying that a gentle breeze sends it, but y’know

    Linux Mint as someone suggested, I’ve ran a long time ago for college on an ancient laptop, and it’s an extreme stable OS, similar to Windows 2000 Pro. I can’t remember it crashing or freezing even once on me, and the Thinkpad T42 has an anemic processor., which I ran with the Conservative Governor