Serving GLM-5.3 at 607.8 tok/s across two Nvidia DGX stations with vLLM

  • Posted 10 hours ago by wondersky
  • 126 points
https://developer.nvidia.com/blog/when-one-gpu-isnt-enough-serving-glm-5-3-across-two-nvidia-dgx-stations-with-vllm/

12 comments

    Loading..
    Loading..
    Loading..
    Loading..
    Loading..
    Loading..
    Loading..
    Loading..
    Loading..
    Loading..
    Loading..
    Loading..