5 ms·

I cannot think why someone would run those workflows on a Windows laptop, unless someone has way too much money to spend.

• bigfishrunning 5mo ago
  > someone has way too much money to spend. that's what nvidia is hoping for
• thinkingtoilet 5mo ago
  If the workload is offloaded to the chip, why would the host platform matter?
  • fg137 5mo ago
    Lots of machine learning workflows support Linux better than Windows, if they run on Windows at all. (e.g. https://docs.vllm.ai/en/latest/getting_started/quickstart/ https://docs.vllm.ai/en/latest/getting_started/quickstart/ ) DGX Spark runs Linux, and nobody is going to install Windows on that machine. This laptop got it backwards. If someone decides to run Ollama for local inference with this laptop, they fit perfectly into the "has too much money to waste" bracket, which is addressed by a few other comments in the discussion.
    • chris_money202 5mo ago
      WSL
      • rvz 5mo ago
        Believe it or not, Windows (WSL) is the best Linux distro and Nvidia knows that.
      • fg137 5mo ago
        It often works, but you always lose something compared to native Linux.
        • pjmlp 5mo ago
          Nah, before WSL I was already using a mix of Virtual Box and VMware Workstation, between home and work computers. Installing Linux natively on laptops has always had some specific features not working. Even my Asus netbook, which came with Linux pre-installed, had wlan issues that I learned to work around with, and the driver never supported the same OpenGL version as the Windows one (3.3 vs 4.1).
          • fg137 4mo ago
            My comment was saying you lose something (e.g. performance) when using WSL2 compared to native Linux on a proper workstation. Linux driver has always been an issue on laptops, but that's not the concern for running Python code.
    • lostmsu 5mo ago
      vllm-windows works well enough
    • woctordho 5mo ago
      There is vllm-windows, and it's just as fast as on Linux. BTW I'm the maintainer of triton-windows.